Propositionalization: Robot sensor data¶

In this notebook, we compare getML's FastProp against well-known feature engineering libraries featuretools and tsfresh.

Summary:

Prediction type: Regression
Domain: Robotics
Prediction target: The force vector on the robot's arm
Population size: 15001

Background¶

A common approach to feature engineering is to generate attribute-value representations from relational data by applying a fixed set of aggregations to columns of interest and perform a feature selection on the (possibly large) set of generated features afterwards. In academia, this approach is called propositionalization.

getML's FastProp is an implementation of this propositionalization approach that has been optimized for speed and memory efficiency. In this notebook, we want to demonstrate how – well – fast FastProp is. To this end, we will benchmark FastProp against the popular feature engineering libraries featuretools and tsfresh. Both of these libraries use propositionalization approaches for feature engineering.

The data set has been generously provided by Erik Berger who originally collected it for his dissertation:

Berger, E. (2018). Behavior-Specific Proprioception Models for Robotic Force Estimation: A Machine Learning Approach. Freiberg, Germany: Technische Universitaet Bergakademie Freiberg.

We begin by importing the libraries and setting the project.

In [1]:

  Copied!     
 
%pip install -q "getml==1.5.0" "featuretools==1.31.0" "tsfresh==0.20.3"
%pip install -q "getml==1.5.0" "featuretools==1.31.0" "tsfresh==0.20.3"

Note: you may need to restart the kernel to use updated packages.

In [2]:

  Copied!     
 
import os
import sys

os.environ["PYARROW_IGNORE_TIMEZONE"] = "1"
from pathlib import Path

import numpy as np
import pandas as pd
import getml

print(f"getML API version: {getml.__version__}\n")
import os import sys os.environ["PYARROW_IGNORE_TIMEZONE"] = "1" from pathlib import Path import numpy as np import pandas as pd import getml print(f"getML API version: {getml.__version__}\n")

getML API version: 1.5.0

In [3]:

  Copied!     
 
getml.engine.launch(allow_remote_ips=True, token="token")
getml.engine.set_project("robot")
getml.engine.launch(allow_remote_ips=True, token="token") getml.engine.set_project("robot")

getML Engine is already running.

Connected to project 'robot'.

In [4]:

  Copied!     
 
# If we are in Colab, we need to fetch the utils folder from the repository
if os.getenv("COLAB_RELEASE_TAG"):
    !curl -L https://api.github.com/repos/getml/getml-demo/tarball/master | tar --wildcards --strip-components=1 -xz '*utils*'
# If we are in Colab, we need to fetch the utils folder from the repository if os.getenv("COLAB_RELEASE_TAG"): !curl -L https://api.github.com/repos/getml/getml-demo/tarball/master | tar --wildcards --strip-components=1 -xz '*utils*'

In [5]:

  Copied!     
 
parent = Path(os.getcwd()).parent.as_posix()

if parent not in sys.path:
    sys.path.append(parent)

from utils import Benchmark, FTTimeSeriesBuilder, TSFreshBuilder
parent = Path(os.getcwd()).parent.as_posix() if parent not in sys.path: sys.path.append(parent) from utils import Benchmark, FTTimeSeriesBuilder, TSFreshBuilder

1. Loading data¶

1.1 Download from source¶

In [6]:

  Copied!     
 
data_all = getml.data.DataFrame.from_csv(
    "https://static.getml.com/datasets/robotarm/robot-demo.csv", "data_all"
)
data_all = getml.data.DataFrame.from_csv( "https://static.getml.com/datasets/robotarm/robot-demo.csv", "data_all" )

Downloading https://static.getml.com/datasets/robotarm/robot-demo.csv to 
/tmp/getml/static.getml.com/datasets/robotarm/robot-demo.csv...

  Downloading robot-demo.csv... ━━━━━━━━━━━━━━━━━━━━ 100% • 14.7/14.7 MB • 00:00

In [7]:

  Copied!     
 
data_all
data_all

Out[7]:

name	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	98	99	100	101	102	103	104	105	106	f_x	f_y	f_z
role	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float
0	3.4098	-0.3274	0.9604	-3.7436	-1.0191	-6.0205	0	0	0	0	0	0	0	0	0	0	0	0	8.38e-17	-4.8116	-1.4033	-0.1369	0.002472	0	9.803e-16	-55.642	-16.312	-1.2042	0.02167	0	3.4098	-0.3274	0.9605	-3.7437	-1.0191	-6.0205	0	0	0	0	0	0	0.1233	-6.5483	-2.8045	-0.8296	0.07625	-0.1906	0.1211	-6.5483	-2.8157	-0.8281	0.07015	-0.1983	0.7699	0.41	0.08279	-1.4094	0.786	-0.3682	0	0	0	0	0	0	-22.654	-11.503	-18.673	-3.5155	5.8354	-2.05	0.7699	0.41	0.08278	-1.4094	0.786	-0.3681	0	0	0	0	0	0	48.069	48.009	0.9668	47.834	47.925	47.818	47.834	47.955	47.971	-11.03	6.9	-7.33
1	3.4098	-0.3274	0.9604	-3.7436	-1.0191	-6.0205	0	0	0	0	0	0	0	0	0	0	0	0	8.38e-17	-4.8116	-1.4033	-0.1369	0.002472	0	9.803e-16	-55.642	-16.312	-1.2042	0.02167	0	3.4098	-0.3274	0.9604	-3.7437	-1.0191	-6.0205	0	0	0	0	0	0	0.1188	-6.5506	-2.8404	-0.8281	0.06405	-0.1998	0.1211	-6.5483	-2.8157	-0.8281	0.07015	-0.1983	0.7699	0.41	0.0828	-1.4094	0.7859	-0.3682	0	0	0	0	0	0	-21.627	-11.046	-18.66	-3.5395	5.7577	-1.9805	0.7699	0.41	0.08278	-1.4094	0.786	-0.3681	0	0	0	0	0	0	48.009	48.009	0.8594	47.834	47.925	47.818	47.834	47.955	47.971	-10.848	6.7218	-7.4427
2	3.4098	-0.3274	0.9604	-3.7436	-1.0191	-6.0205	0	0	0	0	0	0	0	0	0	0	0	0	8.38e-17	-4.8116	-1.4033	-0.1369	0.002472	0	9.803e-16	-55.642	-16.312	-1.2042	0.02167	0	3.4098	-0.3274	0.9605	-3.7437	-1.0191	-6.0205	0	0	0	0	0	0	0.1099	-6.5438	-2.8	-0.8205	0.07473	-0.183	0.1211	-6.5483	-2.8157	-0.8281	0.07015	-0.1922	0.7699	0.41	0.08279	-1.4094	0.7859	-0.3682	0	0	0	0	0	0	-23.843	-12.127	-18.393	-3.6453	5.978	-1.9978	0.7699	0.41	0.08278	-1.4094	0.786	-0.3681	0	0	0	0	0	0	48.009	48.069	0.931	47.879	47.925	47.818	47.834	47.955	47.971	-10.666	6.5436	-7.5555
3	3.4098	-0.3274	0.9604	-3.7436	-1.0191	-6.0205	0	0	0	0	0	0	0	0	0	0	0	0	8.38e-17	-4.8116	-1.4033	-0.1369	0.002472	0	9.803e-16	-55.642	-16.312	-1.2042	0.02167	0	3.4098	-0.3273	0.9604	-3.7437	-1.0191	-6.0205	0	0	0	0	0	0	0.1233	-6.5483	-2.8224	-0.8266	0.07168	-0.1998	0.1211	-6.5483	-2.8157	-0.8281	0.07015	-0.1967	0.7699	0.41	0.08275	-1.4094	0.786	-0.3681	0	0	0	0	0	0	-21.772	-10.872	-18.691	-3.5512	5.6648	-1.9976	0.7699	0.41	0.08278	-1.4094	0.786	-0.3681	0	0	0	0	0	0	48.069	48.069	0.931	47.879	47.925	47.818	47.834	47.955	47.971	-10.507	6.4533	-7.65
4	3.4098	-0.3274	0.9604	-3.7436	-1.0191	-6.0205	0	0	0	0	0	0	0	0	0	0	0	0	8.38e-17	-4.8116	-1.4033	-0.1369	0.002472	0	9.803e-16	-55.642	-16.312	-1.2042	0.02167	0	3.4098	-0.3274	0.9604	-3.7437	-1.0191	-6.0205	0	0	0	0	0	0	0.1255	-6.5394	-2.8	-0.8327	0.07473	-0.1952	0.1211	-6.5483	-2.8157	-0.8327	0.07015	-0.1922	0.7699	0.41	0.08278	-1.4094	0.786	-0.3681	0	0	0	0	0	0	-22.823	-11.645	-18.524	-3.5305	5.8712	-2.0096	0.7699	0.41	0.08278	-1.4094	0.786	-0.3681	0	0	0	0	0	0	48.069	48.069	0.8952	47.879	47.925	47.818	47.834	47.955	47.971	-10.413	6.6267	-7.69
	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
14996	3.0837	-0.8836	1.4501	-2.2102	-1.559	-5.3265	-0.03151	-0.05375	0.04732	0.1482	-0.05218	0.06706	0.2969	0.5065	-0.4459	-1.3963	0.4916	-0.6319	-0.3694	-4.1879	-1.1847	-0.09441	-0.1568	0.1898	1.1605	-42.951	-19.023	-2.6343	0.1551	-0.1338	3.0836	-0.8836	1.4503	-2.2101	-1.5591	-5.3263	-0.03347	-0.05585	0.04805	0.151	-0.05513	0.07114	-0.3564	-6.0394	-2.3001	-0.2181	-0.1159	0.09608	-0.3632	-6.0394	-2.3023	-0.212	-0.125	0.1113	0.7116	0.06957	0.06036	-0.8506	2.9515	-0.03352	-0.03558	-0.03029	0.002444	-0.04208	0.1458	-0.1098	-0.8784	-0.07291	-37.584	0.0001132	-2.1031	0.03318	0.7117	0.0697	0.06044	-0.8511	2.951	-0.03356	-0.03508	-0.02849	0.001571	-0.03951	0.1442	-0.1036	48.069	48.009	0.8952	47.818	47.834	47.818	47.803	47.94	47.94	10.84	-1.41	16.14
14997	3.0835	-0.884	1.4505	-2.2091	-1.5594	-5.326	-0.02913	-0.0497	0.04376	0.137	-0.04825	0.062	0.2969	0.5065	-0.4459	-1.3963	0.4916	-0.6319	-0.3677	-4.1837	-1.1874	-0.09682	-0.1562	0.189	1.1592	-42.937	-19.023	-2.6331	0.1545	-0.1338	3.0833	-0.8841	1.4507	-2.209	-1.5596	-5.3258	-0.02909	-0.04989	0.04198	0.1481	-0.05465	0.06249	-0.3161	-6.1179	-2.253	-0.3752	-0.03965	0.08693	-0.3273	-6.1022	-2.2597	-0.366	-0.05033	0.0915	0.7114	0.06932	0.06039	-0.8497	2.953	-0.03359	-0.0335	-0.02723	0.001208	-0.04242	0.1428	-0.0967	-2.7137	0.8552	-38.514	-0.6088	-3.2383	-0.9666	0.7114	0.06948	0.06045	-0.8503	2.9525	-0.03359	-0.03246	-0.02633	0.001469	-0.03657	0.1333	-0.09571	48.009	48.009	0.8594	47.818	47.834	47.818	47.803	47.94	47.94	10.857	-1.52	15.943
14998	3.0833	-0.8844	1.4508	-2.208	-1.5598	-5.3256	-0.02676	-0.04565	0.04019	0.1258	-0.04431	0.05695	0.2969	0.5065	-0.4459	-1.3963	0.4916	-0.6319	-0.3659	-4.1797	-1.1901	-0.09922	-0.1555	0.1881	1.1579	-42.924	-19.023	-2.6321	0.154	-0.1338	3.0831	-0.8844	1.451	-2.2078	-1.56	-5.3253	-0.02776	-0.04382	0.03652	0.1295	-0.05064	0.04818	-0.343	-6.2569	-2.1566	-0.3035	0.00305	0.1434	-0.3385	-6.2322	-2.1589	-0.302	-0.00915	0.1571	0.7111	0.06912	0.06039	-0.849	2.9544	-0.0337	-0.02911	-0.02589	0.001292	-0.04046	0.1246	-0.08058	4.2749	1.0128	-36.412	-1.2811	-0.4296	-1.1013	0.7112	0.06928	0.06046	-0.8495	2.9538	-0.03362	-0.02984	-0.02417	0.001364	-0.03362	0.1224	-0.08786	48.009	48.009	0.931	47.818	47.834	47.818	47.803	47.94	47.94	10.89	-1.74	15.55
14999	3.0831	-0.8847	1.4511	-2.2071	-1.5602	-5.3251	-0.02438	-0.0416	0.03662	0.1147	-0.04038	0.0519	0.2969	0.5065	-0.4459	-1.3963	0.4916	-0.6319	-0.3642	-4.1758	-1.1928	-0.1016	-0.1548	0.1873	1.1568	-42.912	-19.023	-2.6311	0.1535	-0.1338	3.0829	-0.8848	1.4513	-2.2068	-1.5604	-5.3249	-0.02149	-0.04059	0.03417	0.1202	-0.0395	0.04178	-0.4237	-6.2703	-2.0939	-0.302	-0.01372	0.1739	-0.4125	-6.2569	-2.0916	-0.2943	-0.02898	0.1891	0.7109	0.06894	0.06039	-0.8484	2.9557	-0.03384	-0.02738	-0.01982	0.001031	-0.03028	0.1157	-0.06702	11.518	1.5002	-39.314	-1.8671	-0.3734	-0.5733	0.7109	0.06909	0.06047	-0.8488	2.955	-0.03364	-0.02721	-0.02201	0.001255	-0.03067	0.1115	-0.08003	48.009	48.009	0.931	47.818	47.834	47.818	47.803	47.94	47.94	11.29	-1.4601	15.743
15000	3.0829	-0.885	1.4514	-2.2062	-1.5605	-5.3247	-0.02201	-0.03755	0.03305	0.1035	-0.03645	0.04684	0.2969	0.5065	-0.4459	-1.3963	0.4916	-0.6319	-0.3624	-4.172	-1.1955	-0.1041	-0.1542	0.1864	1.1558	-42.901	-19.023	-2.6302	0.1531	-0.1338	3.0827	-0.8851	1.4516	-2.2059	-1.5607	-5.3246	-0.02096	-0.03808	0.02958	0.1171	-0.03289	0.03883	-0.417	-6.2434	-2.058	-0.4102	-0.04728	0.1967	-0.4237	-6.2367	-2.0714	-0.4163	-0.0671	0.2059	0.7107	0.06878	0.06041	-0.8478	2.9567	-0.03382	-0.02535	-0.01854	0.001614	-0.02421	0.11	-0.06304	15.099	2.936	-39.068	-1.9402	0.139	-0.2674	0.7107	0.06893	0.06048	-0.8482	2.9561	-0.03367	-0.02458	-0.01986	0.001142	-0.0277	0.1007	-0.07221	48.009	48.069	0.8952	47.818	47.834	47.818	47.803	47.94	47.955	11.69	-1.1801	15.937

15001 rows x 96 columns
memory usage: 11.52 MB
name: data_all
type: getml.DataFrame

1.2 Prepare data¶

The force vector consists of three component (f_x, f_y and f_z), meaning that we have three targets. For this comparison, we only predict the first component (f_x).

Also, we want to speed things up a little, so we only use 10 columns. A previous analysis has revealed that the predictive power is mainly extracted from these 10 columns:

In [8]:

  Copied!     
 
only_use = ["30", "34", "37", "38", "4", "59", "61", "7", "77", "78"]

data_all.set_role(["f_x"], getml.data.roles.target)
data_all.set_role(only_use, getml.data.roles.numerical)
only_use = ["30", "34", "37", "38", "4", "59", "61", "7", "77", "78"] data_all.set_role(["f_x"], getml.data.roles.target) data_all.set_role(only_use, getml.data.roles.numerical)

This is what the data set looks like:

In [9]:

  Copied!     
 
data_all
data_all

Out[9]:

name	f_x	30	34	37	38	4	59	61	7	77	78	3	5	6	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	31	32	33	35	36	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	60	62	63	64	65	66	67	68	69	70	71	72	73	74	75	76	79	80	81	82	83	84	85	86	98	99	100	101	102	103	104	105	106	f_y	f_z
role	target	numerical	numerical	numerical	numerical	numerical	numerical	numerical	numerical	numerical	numerical	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float	unused_float
0	-11.03	-1.2042	-0.3274	-1.0191	-6.0205	-0.3274	0.08279	0.786	-1.0191	0.08278	-1.4094	3.4098	0.9604	-3.7436	-6.0205	0	0	0	0	0	0	0	0	0	0	0	0	8.38e-17	-4.8116	-1.4033	-0.1369	0.002472	0	9.803e-16	-55.642	-16.312	0.02167	0	3.4098	0.9605	-3.7437	0	0	0	0	0	0	0.1233	-6.5483	-2.8045	-0.8296	0.07625	-0.1906	0.1211	-6.5483	-2.8157	-0.8281	0.07015	-0.1983	0.7699	0.41	-1.4094	-0.3682	0	0	0	0	0	0	-22.654	-11.503	-18.673	-3.5155	5.8354	-2.05	0.7699	0.41	0.786	-0.3681	0	0	0	0	0	0	48.069	48.009	0.9668	47.834	47.925	47.818	47.834	47.955	47.971	6.9	-7.33
1	-10.848	-1.2042	-0.3274	-1.0191	-6.0205	-0.3274	0.0828	0.7859	-1.0191	0.08278	-1.4094	3.4098	0.9604	-3.7436	-6.0205	0	0	0	0	0	0	0	0	0	0	0	0	8.38e-17	-4.8116	-1.4033	-0.1369	0.002472	0	9.803e-16	-55.642	-16.312	0.02167	0	3.4098	0.9604	-3.7437	0	0	0	0	0	0	0.1188	-6.5506	-2.8404	-0.8281	0.06405	-0.1998	0.1211	-6.5483	-2.8157	-0.8281	0.07015	-0.1983	0.7699	0.41	-1.4094	-0.3682	0	0	0	0	0	0	-21.627	-11.046	-18.66	-3.5395	5.7577	-1.9805	0.7699	0.41	0.786	-0.3681	0	0	0	0	0	0	48.009	48.009	0.8594	47.834	47.925	47.818	47.834	47.955	47.971	6.7218	-7.4427
2	-10.666	-1.2042	-0.3274	-1.0191	-6.0205	-0.3274	0.08279	0.7859	-1.0191	0.08278	-1.4094	3.4098	0.9604	-3.7436	-6.0205	0	0	0	0	0	0	0	0	0	0	0	0	8.38e-17	-4.8116	-1.4033	-0.1369	0.002472	0	9.803e-16	-55.642	-16.312	0.02167	0	3.4098	0.9605	-3.7437	0	0	0	0	0	0	0.1099	-6.5438	-2.8	-0.8205	0.07473	-0.183	0.1211	-6.5483	-2.8157	-0.8281	0.07015	-0.1922	0.7699	0.41	-1.4094	-0.3682	0	0	0	0	0	0	-23.843	-12.127	-18.393	-3.6453	5.978	-1.9978	0.7699	0.41	0.786	-0.3681	0	0	0	0	0	0	48.009	48.069	0.931	47.879	47.925	47.818	47.834	47.955	47.971	6.5436	-7.5555
3	-10.507	-1.2042	-0.3273	-1.0191	-6.0205	-0.3274	0.08275	0.786	-1.0191	0.08278	-1.4094	3.4098	0.9604	-3.7436	-6.0205	0	0	0	0	0	0	0	0	0	0	0	0	8.38e-17	-4.8116	-1.4033	-0.1369	0.002472	0	9.803e-16	-55.642	-16.312	0.02167	0	3.4098	0.9604	-3.7437	0	0	0	0	0	0	0.1233	-6.5483	-2.8224	-0.8266	0.07168	-0.1998	0.1211	-6.5483	-2.8157	-0.8281	0.07015	-0.1967	0.7699	0.41	-1.4094	-0.3681	0	0	0	0	0	0	-21.772	-10.872	-18.691	-3.5512	5.6648	-1.9976	0.7699	0.41	0.786	-0.3681	0	0	0	0	0	0	48.069	48.069	0.931	47.879	47.925	47.818	47.834	47.955	47.971	6.4533	-7.65
4	-10.413	-1.2042	-0.3274	-1.0191	-6.0205	-0.3274	0.08278	0.786	-1.0191	0.08278	-1.4094	3.4098	0.9604	-3.7436	-6.0205	0	0	0	0	0	0	0	0	0	0	0	0	8.38e-17	-4.8116	-1.4033	-0.1369	0.002472	0	9.803e-16	-55.642	-16.312	0.02167	0	3.4098	0.9604	-3.7437	0	0	0	0	0	0	0.1255	-6.5394	-2.8	-0.8327	0.07473	-0.1952	0.1211	-6.5483	-2.8157	-0.8327	0.07015	-0.1922	0.7699	0.41	-1.4094	-0.3681	0	0	0	0	0	0	-22.823	-11.645	-18.524	-3.5305	5.8712	-2.0096	0.7699	0.41	0.786	-0.3681	0	0	0	0	0	0	48.069	48.069	0.8952	47.879	47.925	47.818	47.834	47.955	47.971	6.6267	-7.69
	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
14996	10.84	-2.6343	-0.8836	-1.5591	-5.3263	-0.8836	0.06036	2.9515	-1.559	0.06044	-0.8511	3.0837	1.4501	-2.2102	-5.3265	-0.03151	-0.05375	0.04732	0.1482	-0.05218	0.06706	0.2969	0.5065	-0.4459	-1.3963	0.4916	-0.6319	-0.3694	-4.1879	-1.1847	-0.09441	-0.1568	0.1898	1.1605	-42.951	-19.023	0.1551	-0.1338	3.0836	1.4503	-2.2101	-0.03347	-0.05585	0.04805	0.151	-0.05513	0.07114	-0.3564	-6.0394	-2.3001	-0.2181	-0.1159	0.09608	-0.3632	-6.0394	-2.3023	-0.212	-0.125	0.1113	0.7116	0.06957	-0.8506	-0.03352	-0.03558	-0.03029	0.002444	-0.04208	0.1458	-0.1098	-0.8784	-0.07291	-37.584	0.0001132	-2.1031	0.03318	0.7117	0.0697	2.951	-0.03356	-0.03508	-0.02849	0.001571	-0.03951	0.1442	-0.1036	48.069	48.009	0.8952	47.818	47.834	47.818	47.803	47.94	47.94	-1.41	16.14
14997	10.857	-2.6331	-0.8841	-1.5596	-5.3258	-0.884	0.06039	2.953	-1.5594	0.06045	-0.8503	3.0835	1.4505	-2.2091	-5.326	-0.02913	-0.0497	0.04376	0.137	-0.04825	0.062	0.2969	0.5065	-0.4459	-1.3963	0.4916	-0.6319	-0.3677	-4.1837	-1.1874	-0.09682	-0.1562	0.189	1.1592	-42.937	-19.023	0.1545	-0.1338	3.0833	1.4507	-2.209	-0.02909	-0.04989	0.04198	0.1481	-0.05465	0.06249	-0.3161	-6.1179	-2.253	-0.3752	-0.03965	0.08693	-0.3273	-6.1022	-2.2597	-0.366	-0.05033	0.0915	0.7114	0.06932	-0.8497	-0.03359	-0.0335	-0.02723	0.001208	-0.04242	0.1428	-0.0967	-2.7137	0.8552	-38.514	-0.6088	-3.2383	-0.9666	0.7114	0.06948	2.9525	-0.03359	-0.03246	-0.02633	0.001469	-0.03657	0.1333	-0.09571	48.009	48.009	0.8594	47.818	47.834	47.818	47.803	47.94	47.94	-1.52	15.943
14998	10.89	-2.6321	-0.8844	-1.56	-5.3253	-0.8844	0.06039	2.9544	-1.5598	0.06046	-0.8495	3.0833	1.4508	-2.208	-5.3256	-0.02676	-0.04565	0.04019	0.1258	-0.04431	0.05695	0.2969	0.5065	-0.4459	-1.3963	0.4916	-0.6319	-0.3659	-4.1797	-1.1901	-0.09922	-0.1555	0.1881	1.1579	-42.924	-19.023	0.154	-0.1338	3.0831	1.451	-2.2078	-0.02776	-0.04382	0.03652	0.1295	-0.05064	0.04818	-0.343	-6.2569	-2.1566	-0.3035	0.00305	0.1434	-0.3385	-6.2322	-2.1589	-0.302	-0.00915	0.1571	0.7111	0.06912	-0.849	-0.0337	-0.02911	-0.02589	0.001292	-0.04046	0.1246	-0.08058	4.2749	1.0128	-36.412	-1.2811	-0.4296	-1.1013	0.7112	0.06928	2.9538	-0.03362	-0.02984	-0.02417	0.001364	-0.03362	0.1224	-0.08786	48.009	48.009	0.931	47.818	47.834	47.818	47.803	47.94	47.94	-1.74	15.55
14999	11.29	-2.6311	-0.8848	-1.5604	-5.3249	-0.8847	0.06039	2.9557	-1.5602	0.06047	-0.8488	3.0831	1.4511	-2.2071	-5.3251	-0.02438	-0.0416	0.03662	0.1147	-0.04038	0.0519	0.2969	0.5065	-0.4459	-1.3963	0.4916	-0.6319	-0.3642	-4.1758	-1.1928	-0.1016	-0.1548	0.1873	1.1568	-42.912	-19.023	0.1535	-0.1338	3.0829	1.4513	-2.2068	-0.02149	-0.04059	0.03417	0.1202	-0.0395	0.04178	-0.4237	-6.2703	-2.0939	-0.302	-0.01372	0.1739	-0.4125	-6.2569	-2.0916	-0.2943	-0.02898	0.1891	0.7109	0.06894	-0.8484	-0.03384	-0.02738	-0.01982	0.001031	-0.03028	0.1157	-0.06702	11.518	1.5002	-39.314	-1.8671	-0.3734	-0.5733	0.7109	0.06909	2.955	-0.03364	-0.02721	-0.02201	0.001255	-0.03067	0.1115	-0.08003	48.009	48.009	0.931	47.818	47.834	47.818	47.803	47.94	47.94	-1.4601	15.743
15000	11.69	-2.6302	-0.8851	-1.5607	-5.3246	-0.885	0.06041	2.9567	-1.5605	0.06048	-0.8482	3.0829	1.4514	-2.2062	-5.3247	-0.02201	-0.03755	0.03305	0.1035	-0.03645	0.04684	0.2969	0.5065	-0.4459	-1.3963	0.4916	-0.6319	-0.3624	-4.172	-1.1955	-0.1041	-0.1542	0.1864	1.1558	-42.901	-19.023	0.1531	-0.1338	3.0827	1.4516	-2.2059	-0.02096	-0.03808	0.02958	0.1171	-0.03289	0.03883	-0.417	-6.2434	-2.058	-0.4102	-0.04728	0.1967	-0.4237	-6.2367	-2.0714	-0.4163	-0.0671	0.2059	0.7107	0.06878	-0.8478	-0.03382	-0.02535	-0.01854	0.001614	-0.02421	0.11	-0.06304	15.099	2.936	-39.068	-1.9402	0.139	-0.2674	0.7107	0.06893	2.9561	-0.03367	-0.02458	-0.01986	0.001142	-0.0277	0.1007	-0.07221	48.009	48.069	0.8952	47.818	47.834	47.818	47.803	47.94	47.955	-1.1801	15.937

15001 rows x 96 columns
memory usage: 11.52 MB
name: data_all
type: getml.DataFrame

1.3 Separate data into a training and testing set¶

We also want to separate the data set into a training and testing set. We do so by using the first 10,500 measurements for training and then using the remainder for testing.

In [10]:

  Copied!     
 
split = getml.data.split.time(data_all, "rowid", test=10500)
split
split = getml.data.split.time(data_all, "rowid", test=10500) split

Out[10]:


0	train
1	train
2	train
3	train
4	train
	...

unknown number of rows
type: StringColumnView

In [11]:

  Copied!     
 
time_series = getml.data.TimeSeries(
    population=data_all,
    split=split,
    time_stamps="rowid",
    lagged_targets=False,
    memory=30,
)

time_series
time_series = getml.data.TimeSeries( population=data_all, split=split, time_stamps="rowid", lagged_targets=False, memory=30, ) time_series

Out[11]:

data model

diagram

staging

	data frames	staging table
0	population	POPULATION__STAGING_TABLE_1
1	data_all	DATA_ALL__STAGING_TABLE_2

container

population

	subset	name	rows	type
0	test	data_all	unknown	View
1	train	data_all	unknown	View

peripheral

	name	rows	type
0	data_all	15001	View

2. Predictive modeling¶

2.1 Propositionalization with getML's FastProp¶

In [12]:

  Copied!     
 
fast_prop = getml.feature_learning.FastProp(
    loss_function=getml.feature_learning.loss_functions.SquareLoss,
)
fast_prop = getml.feature_learning.FastProp( loss_function=getml.feature_learning.loss_functions.SquareLoss, )

In [13]:

  Copied!     
 
pipe_fp_fl = getml.pipeline.Pipeline(
    data_model=time_series.data_model,
    feature_learners=[fast_prop],
    tags=["feature learning", "fastprop"],
)
pipe_fp_fl = getml.pipeline.Pipeline( data_model=time_series.data_model, feature_learners=[fast_prop], tags=["feature learning", "fastprop"], )

In [14]:

  Copied!     
 
pipe_fp_fl.check(time_series.train)
pipe_fp_fl.check(time_series.train)

Checking data model...

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  Checking... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

OK.

In [15]:

  Copied!     
 
benchmark = Benchmark()
benchmark = Benchmark()

In [16]:

  Copied!     
 
with benchmark("fastprop"):
    pipe_fp_fl.fit(time_series.train)
    fastprop_train = pipe_fp_fl.transform(time_series.train, df_name="fastprop_train")
with benchmark("fastprop"): pipe_fp_fl.fit(time_series.train) fastprop_train = pipe_fp_fl.transform(time_series.train, df_name="fastprop_train")

Checking data model...

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

OK.

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  FastProp: Trying 134 features... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

Trained pipeline.

Time taken: 0:00:00.032192.

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  Preprocessing... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  FastProp: Building features... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

In [17]:

  Copied!     
 
fastprop_test = pipe_fp_fl.transform(time_series.test, df_name="fastprop_test")
fastprop_test = pipe_fp_fl.transform(time_series.test, df_name="fastprop_test")

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  Preprocessing... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  FastProp: Building features... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

In [18]:

  Copied!     
 
predictor = getml.predictors.XGBoostRegressor()

pipe_fp_pr = getml.pipeline.Pipeline(
    tags=["prediction", "fastprop"], predictors=[predictor]
)
predictor = getml.predictors.XGBoostRegressor() pipe_fp_pr = getml.pipeline.Pipeline( tags=["prediction", "fastprop"], predictors=[predictor] )

In [19]:

  Copied!     
 
pipe_fp_pr.check(fastprop_train)
pipe_fp_pr.check(fastprop_train)

Checking data model...

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  Checking... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

The pipeline check generated 0 issues labeled INFO and 5 issues labeled WARNING.

Out[19]:

	type	label	message
0	WARNING	COLUMN SHOULD BE UNUSED	All non-NULL entries in column 'feature_1_121' in POPULATION__STAGING_TABLE_1 are equal to each other. You should consider setting its role to unused_float or using it for comparison only (you can do the latter by setting a unit that contains 'comparison only').
1	WARNING	COLUMN SHOULD BE UNUSED	All non-NULL entries in column 'feature_1_126' in POPULATION__STAGING_TABLE_1 are equal to each other. You should consider setting its role to unused_float or using it for comparison only (you can do the latter by setting a unit that contains 'comparison only').
2	WARNING	COLUMN SHOULD BE UNUSED	All non-NULL entries in column 'feature_1_127' in POPULATION__STAGING_TABLE_1 are equal to each other. You should consider setting its role to unused_float or using it for comparison only (you can do the latter by setting a unit that contains 'comparison only').
3	WARNING	COLUMN SHOULD BE UNUSED	All non-NULL entries in column 'feature_1_128' in POPULATION__STAGING_TABLE_1 are equal to each other. You should consider setting its role to unused_float or using it for comparison only (you can do the latter by setting a unit that contains 'comparison only').
4	WARNING	COLUMN SHOULD BE UNUSED	All non-NULL entries in column 'feature_1_132' in POPULATION__STAGING_TABLE_1 are equal to each other. You should consider setting its role to unused_float or using it for comparison only (you can do the latter by setting a unit that contains 'comparison only').

In [20]:

  Copied!     
 
pipe_fp_pr.fit(fastprop_train)
pipe_fp_pr.fit(fastprop_train)

Checking data model...

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

The pipeline check generated 0 issues labeled INFO and 5 issues labeled WARNING.

To see the issues in full, run .check() on the pipeline.

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  XGBoost: Training as predictor... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:04

Trained pipeline.

Time taken: 0:00:04.112529.

Out[20]:

Pipeline(data_model='population',
         feature_learners=[],
         feature_selectors=[],
         include_categorical=False,
         loss_function='SquareLoss',
         peripheral=[],
         predictors=['XGBoostRegressor'],
         preprocessors=[],
         share_selected_features=0.5,
         tags=['prediction', 'fastprop'])

In [21]:

  Copied!     
 
pipe_fp_pr.score(fastprop_test)
pipe_fp_pr.score(fastprop_test)

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  Preprocessing... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

Out[21]:

	date time	set used	target	mae	rmse	rsquared
0	2024-09-13 14:16:39	fastprop_train	f_x	0.4383	0.5764	0.9963
1	2024-09-13 14:16:39	fastprop_test	f_x	0.5516	0.7237	0.9951

2.2 Propositionalization with featuretools¶

In [22]:

  Copied!     
 
data_train = time_series.train.population.to_df("data_train")
data_test = time_series.test.population.to_df("data_test")
data_train = time_series.train.population.to_df("data_train") data_test = time_series.test.population.to_df("data_test")

In [23]:

  Copied!     
 
dfs_pandas = {}

for df in [data_train, data_test, data_all]:
    dfs_pandas[df.name] = df.to_pandas()
    delete_columns = [
        col for col in dfs_pandas[df.name].columns if col not in only_use + ["f_x"]
    ]
    for col in delete_columns:
        del dfs_pandas[df.name][col]
    dfs_pandas[df.name]["id"] = 1
    dfs_pandas[df.name]["ds"] = pd.to_datetime(
        np.arange(0, dfs_pandas[df.name].shape[0]), unit="s"
    )
dfs_pandas = {} for df in [data_train, data_test, data_all]: dfs_pandas[df.name] = df.to_pandas() delete_columns = [ col for col in dfs_pandas[df.name].columns if col not in only_use + ["f_x"] ] for col in delete_columns: del dfs_pandas[df.name][col] dfs_pandas[df.name]["id"] = 1 dfs_pandas[df.name]["ds"] = pd.to_datetime( np.arange(0, dfs_pandas[df.name].shape[0]), unit="s" )

In [24]:

  Copied!     
 
dfs_pandas["data_train"]
dfs_pandas["data_train"]

Out[24]:

	30	34	37	38	4	59	61	7	77	78	f_x	id	ds
0	-1.2042	-0.32739	-1.0191	-6.0205	-0.32737	0.082791	0.78597	-1.0191	0.082782	-1.4094	-11.0300	1	1970-01-01 00:00:00
1	-1.2042	-0.32739	-1.0191	-6.0205	-0.32737	0.082800	0.78592	-1.0191	0.082782	-1.4094	-10.8480	1	1970-01-01 00:00:01
2	-1.2042	-0.32737	-1.0191	-6.0205	-0.32737	0.082786	0.78594	-1.0191	0.082782	-1.4094	-10.6660	1	1970-01-01 00:00:02
3	-1.2042	-0.32734	-1.0191	-6.0205	-0.32737	0.082755	0.78599	-1.0191	0.082782	-1.4094	-10.5070	1	1970-01-01 00:00:03
4	-1.2042	-0.32736	-1.0191	-6.0205	-0.32737	0.082782	0.78597	-1.0191	0.082782	-1.4094	-10.4130	1	1970-01-01 00:00:04
...	...	...	...	...	...	...	...	...	...	...	...	...	...
10495	-1.1446	-0.37311	-1.0486	-5.9532	-0.37326	0.087343	0.90793	-1.0488	0.087468	-1.4162	-9.7673	1	1970-01-01 02:54:55
10496	-1.1349	-0.37103	-1.0472	-5.9564	-0.37108	0.087241	0.90199	-1.0474	0.087274	-1.4160	-9.9200	1	1970-01-01 02:54:56
10497	-1.1255	-0.36889	-1.0458	-5.9596	-0.36896	0.087055	0.89618	-1.0460	0.087082	-1.4158	-9.7743	1	1970-01-01 02:54:57
10498	-1.1163	-0.36680	-1.0444	-5.9627	-0.36689	0.086907	0.89034	-1.0447	0.086893	-1.4155	-8.6109	1	1970-01-01 02:54:58
10499	-1.1072	-0.36477	-1.0430	-5.9657	-0.36487	0.086720	0.88476	-1.0434	0.086706	-1.4153	-8.4345	1	1970-01-01 02:54:59

10500 rows × 13 columns

In [25]:

  Copied!     
 
ft_builder = FTTimeSeriesBuilder(
    num_features=200,
    horizon=pd.Timedelta(seconds=0),
    memory=pd.Timedelta(seconds=15),
    column_id="id",
    time_stamp="ds",
    target="f_x",
)
ft_builder = FTTimeSeriesBuilder( num_features=200, horizon=pd.Timedelta(seconds=0), memory=pd.Timedelta(seconds=15), column_id="id", time_stamp="ds", target="f_x", )

In [26]:

  Copied!     
 
with benchmark("featuretools"):
    featuretools_train = ft_builder.fit(dfs_pandas["data_train"])

featuretools_test = ft_builder.transform(dfs_pandas["data_test"])
with benchmark("featuretools"): featuretools_train = ft_builder.fit(dfs_pandas["data_train"]) featuretools_test = ft_builder.transform(dfs_pandas["data_test"])

featuretools: Trying features...
Selecting the best out of 442 features...
Time taken: 0h:14m:21.609569

In [27]:

  Copied!     
 
featuretools_train
featuretools_train

Out[27]:

	MIN(peripheral.37)	MIN(peripheral.7)	FIRST(peripheral.37)	FIRST(peripheral.7)	MEDIAN(peripheral.37)	MEDIAN(peripheral.7)	MEAN(peripheral.37)	MEAN(peripheral.7)	SUM(peripheral.37)	SUM(peripheral.7)	...	NUM_ZERO_CROSSINGS(peripheral.77)	NUM_ZERO_CROSSINGS(peripheral.59)	IS_MONOTONICALLY_DECREASING(peripheral.77)	TIME_SINCE_LAST_MAX(peripheral.ds, 4)	TIME_SINCE_LAST_MAX(peripheral.ds, 30)	TIME_SINCE_LAST_MAX(peripheral.ds, 77)	TIME_SINCE_LAST_MAX(peripheral.ds, 7)	f_x	id	ds
_featuretools_index
0	-1.0191	-1.0191	-1.0191	-1.0191	-1.0191	-1.0191	-1.019100	-1.019100	-1.0191	-1.0191	...	0	0	True	1.726237e+09	1.726237e+09	1.726237e+09	1.726237e+09	-11.0300	1	1970-01-01 00:00:00
1	-1.0191	-1.0191	-1.0191	-1.0191	-1.0191	-1.0191	-1.019100	-1.019100	-2.0382	-2.0382	...	0	0	True	1.726237e+09	1.726237e+09	1.726237e+09	1.726237e+09	-10.8480	1	1970-01-01 00:00:01
2	-1.0191	-1.0191	-1.0191	-1.0191	-1.0191	-1.0191	-1.019100	-1.019100	-3.0573	-3.0573	...	0	0	True	1.726237e+09	1.726237e+09	1.726237e+09	1.726237e+09	-10.6660	1	1970-01-01 00:00:02
3	-1.0191	-1.0191	-1.0191	-1.0191	-1.0191	-1.0191	-1.019100	-1.019100	-4.0764	-4.0764	...	0	0	True	1.726237e+09	1.726237e+09	1.726237e+09	1.726237e+09	-10.5070	1	1970-01-01 00:00:03
4	-1.0191	-1.0191	-1.0191	-1.0191	-1.0191	-1.0191	-1.019100	-1.019100	-5.0955	-5.0955	...	0	0	True	1.726237e+09	1.726237e+09	1.726237e+09	1.726237e+09	-10.4130	1	1970-01-01 00:00:04
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
10495	-1.0716	-1.0721	-1.0716	-1.0721	-1.0596	-1.0596	-1.059793	-1.059927	-15.8969	-15.8989	...	0	0	True	1.726227e+09	1.726227e+09	1.726227e+09	1.726227e+09	-9.7673	1	1970-01-01 02:54:55
10496	-1.0698	-1.0702	-1.0698	-1.0702	-1.0580	-1.0580	-1.058167	-1.058280	-15.8725	-15.8742	...	0	0	True	1.726227e+09	1.726227e+09	1.726227e+09	1.726227e+09	-9.9200	1	1970-01-01 02:54:56
10497	-1.0681	-1.0684	-1.0681	-1.0684	-1.0565	-1.0563	-1.056567	-1.056667	-15.8485	-15.8500	...	0	0	True	1.726227e+09	1.726227e+09	1.726227e+09	1.726227e+09	-9.7743	1	1970-01-01 02:54:57
10498	-1.0662	-1.0665	-1.0662	-1.0665	-1.0549	-1.0548	-1.054987	-1.055087	-15.8248	-15.8263	...	0	0	True	1.726227e+09	1.726227e+09	1.726227e+09	1.726227e+09	-8.6109	1	1970-01-01 02:54:58
10499	-1.0644	-1.0648	-1.0644	-1.0648	-1.0534	-1.0532	-1.053440	-1.053547	-15.8016	-15.8032	...	0	0	True	1.726227e+09	1.726227e+09	1.726227e+09	1.726227e+09	-8.4345	1	1970-01-01 02:54:59

10500 rows × 203 columns

In [28]:

  Copied!     
 
roles = {
    getml.data.roles.target: ["f_x"],
    getml.data.roles.join_key: ["id"],
    getml.data.roles.time_stamp: ["ds"],
}

df_featuretools_train = getml.data.DataFrame.from_pandas(
    featuretools_train, name="featuretools_train", roles=roles
)

df_featuretools_test = getml.data.DataFrame.from_pandas(
    featuretools_test, name="featuretools_test", roles=roles
)
roles = { getml.data.roles.target: ["f_x"], getml.data.roles.join_key: ["id"], getml.data.roles.time_stamp: ["ds"], } df_featuretools_train = getml.data.DataFrame.from_pandas( featuretools_train, name="featuretools_train", roles=roles ) df_featuretools_test = getml.data.DataFrame.from_pandas( featuretools_test, name="featuretools_test", roles=roles )

In [29]:

  Copied!     
 
df_featuretools_train.set_role(
    df_featuretools_train.roles.unused, getml.data.roles.numerical
)

df_featuretools_test.set_role(
    df_featuretools_test.roles.unused, getml.data.roles.numerical
)
df_featuretools_train.set_role( df_featuretools_train.roles.unused, getml.data.roles.numerical ) df_featuretools_test.set_role( df_featuretools_test.roles.unused, getml.data.roles.numerical )

In [30]:

  Copied!     
 
predictor = getml.predictors.XGBoostRegressor()

pipe_ft_pr = getml.pipeline.Pipeline(
    tags=["prediction", "featuretools"], predictors=[predictor]
)

pipe_ft_pr
predictor = getml.predictors.XGBoostRegressor() pipe_ft_pr = getml.pipeline.Pipeline( tags=["prediction", "featuretools"], predictors=[predictor] ) pipe_ft_pr

Out[30]:

Pipeline(data_model='population',
         feature_learners=[],
         feature_selectors=[],
         include_categorical=False,
         loss_function='SquareLoss',
         peripheral=[],
         predictors=['XGBoostRegressor'],
         preprocessors=[],
         share_selected_features=0.5,
         tags=['prediction', 'featuretools'])

In [31]:

  Copied!     
 
pipe_ft_pr.fit(df_featuretools_train)
pipe_ft_pr.fit(df_featuretools_train)

Checking data model...

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  Checking... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

OK.

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  XGBoost: Training as predictor... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:04

Trained pipeline.

Time taken: 0:00:04.476476.

Out[31]:

Pipeline(data_model='population',
         feature_learners=[],
         feature_selectors=[],
         include_categorical=False,
         loss_function='SquareLoss',
         peripheral=[],
         predictors=['XGBoostRegressor'],
         preprocessors=[],
         share_selected_features=0.5,
         tags=['prediction', 'featuretools'])

In [32]:

  Copied!     
 
pipe_ft_pr.score(df_featuretools_test)
pipe_ft_pr.score(df_featuretools_test)

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  Preprocessing... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

Out[32]:

	date time	set used	target	mae	rmse	rsquared
0	2024-09-13 14:37:28	featuretools_train	f_x	0.4396	0.584	0.9962
1	2024-09-13 14:37:28	featuretools_test	f_x	0.5828	0.7595	0.9948

2.3 Propositionalization with tsfresh¶

In [33]:

  Copied!     
 
tsfresh_builder = TSFreshBuilder(
    num_features=200,
    memory=15,
    column_id="id",
    time_stamp="ds",
    target="f_x",
)
tsfresh_builder = TSFreshBuilder( num_features=200, memory=15, column_id="id", time_stamp="ds", target="f_x", )

In [34]:

  Copied!     
 
with benchmark("tsfresh"):
    tsfresh_train = tsfresh_builder.fit(dfs_pandas["data_train"])

tsfresh_test = tsfresh_builder.transform(dfs_pandas["data_test"])
with benchmark("tsfresh"): tsfresh_train = tsfresh_builder.fit(dfs_pandas["data_train"]) tsfresh_test = tsfresh_builder.transform(dfs_pandas["data_test"])

Rolling: 100%|██████████| 40/40 [00:02<00:00, 13.60it/s]
Feature Extraction: 100%|██████████| 40/40 [00:09<00:00,  4.09it/s]
Feature Extraction: 100%|██████████| 40/40 [00:11<00:00,  3.64it/s]

Selecting the best out of 130 features...
Time taken: 0h:0m:26.638195

Rolling: 100%|██████████| 40/40 [00:01<00:00, 37.19it/s]
Feature Extraction: 100%|██████████| 40/40 [00:04<00:00,  9.62it/s]
Feature Extraction: 100%|██████████| 40/40 [00:04<00:00,  9.93it/s]

In [35]:

  Copied!     
 
roles = {
    getml.data.roles.target: ["f_x"],
    getml.data.roles.join_key: ["id"],
    getml.data.roles.time_stamp: ["ds"],
}

df_tsfresh_train = getml.data.DataFrame.from_pandas(
    tsfresh_train, name="tsfresh_train", roles=roles
)

df_tsfresh_test = getml.data.DataFrame.from_pandas(
    tsfresh_test, name="tsfresh_test", roles=roles
)
roles = { getml.data.roles.target: ["f_x"], getml.data.roles.join_key: ["id"], getml.data.roles.time_stamp: ["ds"], } df_tsfresh_train = getml.data.DataFrame.from_pandas( tsfresh_train, name="tsfresh_train", roles=roles ) df_tsfresh_test = getml.data.DataFrame.from_pandas( tsfresh_test, name="tsfresh_test", roles=roles )

In [36]:

  Copied!     
 
df_tsfresh_train.set_role(df_tsfresh_train.roles.unused, getml.data.roles.numerical)

df_tsfresh_test.set_role(df_tsfresh_test.roles.unused, getml.data.roles.numerical)
df_tsfresh_train.set_role(df_tsfresh_train.roles.unused, getml.data.roles.numerical) df_tsfresh_test.set_role(df_tsfresh_test.roles.unused, getml.data.roles.numerical)

In [37]:

  Copied!     
 
pipe_tsf_pr = getml.pipeline.Pipeline(
    tags=["predicition", "tsfresh"], predictors=[predictor]
)

pipe_tsf_pr
pipe_tsf_pr = getml.pipeline.Pipeline( tags=["predicition", "tsfresh"], predictors=[predictor] ) pipe_tsf_pr

Out[37]:

Pipeline(data_model='population',
         feature_learners=[],
         feature_selectors=[],
         include_categorical=False,
         loss_function='SquareLoss',
         peripheral=[],
         predictors=['XGBoostRegressor'],
         preprocessors=[],
         share_selected_features=0.5,
         tags=['predicition', 'tsfresh'])

In [38]:

  Copied!     
 
pipe_tsf_pr.check(df_tsfresh_train)
pipe_tsf_pr.check(df_tsfresh_train)

Checking data model...

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  Checking... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

OK.

In [39]:

  Copied!     
 
pipe_tsf_pr.fit(df_tsfresh_train)
pipe_tsf_pr.fit(df_tsfresh_train)

Checking data model...

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

OK.

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  XGBoost: Training as predictor... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:03

Trained pipeline.

Time taken: 0:00:03.942466.

Out[39]:

Pipeline(data_model='population',
         feature_learners=[],
         feature_selectors=[],
         include_categorical=False,
         loss_function='SquareLoss',
         peripheral=[],
         predictors=['XGBoostRegressor'],
         preprocessors=[],
         share_selected_features=0.5,
         tags=['predicition', 'tsfresh'])

In [40]:

  Copied!     
 
pipe_tsf_pr.score(df_tsfresh_test)
pipe_tsf_pr.score(df_tsfresh_test)

  Staging... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00
  Preprocessing... ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 100% • 00:00

Out[40]:

	date time	set used	target	mae	rmse	rsquared
0	2024-09-13 14:38:12	tsfresh_train	f_x	0.4916	0.6636	0.9951
1	2024-09-13 14:38:12	tsfresh_test	f_x	0.5986	0.7906	0.9938

3. Comparison¶

In [41]:

  Copied!     
 
num_features = dict(
    fastprop=134,
    featuretools=158,
    tsfresh=120,
)

runtime_per_feature = [
    benchmark.runtimes["fastprop"] / num_features["fastprop"],
    benchmark.runtimes["featuretools"] / num_features["featuretools"],
    benchmark.runtimes["tsfresh"] / num_features["tsfresh"],
]

features_per_second = [1.0 / r.total_seconds() for r in runtime_per_feature]

normalized_runtime_per_feature = [
    r / runtime_per_feature[0] for r in runtime_per_feature
]

comparison = pd.DataFrame(
    dict(
        runtime=[
            benchmark.runtimes["fastprop"],
            benchmark.runtimes["featuretools"],
            benchmark.runtimes["tsfresh"],
        ],
        num_features=num_features.values(),
        features_per_second=features_per_second,
        normalized_runtime=[
            1,
            benchmark.runtimes["featuretools"] / benchmark.runtimes["fastprop"],
            benchmark.runtimes["tsfresh"] / benchmark.runtimes["fastprop"],
        ],
        normalized_runtime_per_feature=normalized_runtime_per_feature,
        rsquared=[pipe_fp_pr.rsquared, pipe_ft_pr.rsquared, pipe_tsf_pr.rsquared],
        rmse=[pipe_fp_pr.rmse, pipe_ft_pr.rmse, pipe_tsf_pr.rmse],
        mae=[pipe_fp_pr.mae, pipe_ft_pr.mae, pipe_tsf_pr.mae],
    )
)

comparison.index = ["getML: FastProp", "featuretools", "tsfresh"]
num_features = dict( fastprop=134, featuretools=158, tsfresh=120, ) runtime_per_feature = [ benchmark.runtimes["fastprop"] / num_features["fastprop"], benchmark.runtimes["featuretools"] / num_features["featuretools"], benchmark.runtimes["tsfresh"] / num_features["tsfresh"], ] features_per_second = [1.0 / r.total_seconds() for r in runtime_per_feature] normalized_runtime_per_feature = [ r / runtime_per_feature[0] for r in runtime_per_feature ] comparison = pd.DataFrame( dict( runtime=[ benchmark.runtimes["fastprop"], benchmark.runtimes["featuretools"], benchmark.runtimes["tsfresh"], ], num_features=num_features.values(), features_per_second=features_per_second, normalized_runtime=[ 1, benchmark.runtimes["featuretools"] / benchmark.runtimes["fastprop"], benchmark.runtimes["tsfresh"] / benchmark.runtimes["fastprop"], ], normalized_runtime_per_feature=normalized_runtime_per_feature, rsquared=[pipe_fp_pr.rsquared, pipe_ft_pr.rsquared, pipe_tsf_pr.rsquared], rmse=[pipe_fp_pr.rmse, pipe_ft_pr.rmse, pipe_tsf_pr.rmse], mae=[pipe_fp_pr.mae, pipe_ft_pr.mae, pipe_tsf_pr.mae], ) ) comparison.index = ["getML: FastProp", "featuretools", "tsfresh"]

In [42]:

  Copied!     
 
comparison
comparison

Out[42]:

	runtime	num_features	features_per_second	normalized_runtime	normalized_runtime_per_feature	rsquared	rmse	mae
getML: FastProp	0 days 00:00:00.398347	134	336.360579	1.000000	1.000000	0.995058	0.723716	0.551608
featuretools	0 days 00:14:21.611109	158	0.183377	2162.966230	1834.253280	0.994784	0.759460	0.582831
tsfresh	0 days 00:00:26.638362	120	4.504789	66.872255	74.667339	0.993836	0.790602	0.598600

In [43]:

  Copied!     
 
comparison.to_csv("comparisons/robot.csv")
comparison.to_csv("comparisons/robot.csv")

In [ ]:

  Copied!     
 
getml.engine.shutdown()
getml.engine.shutdown()