Tutorial on fault diagnosis task

import numpy as np
from sklearn.preprocessing import StandardScaler

from ice.fault_diagnosis.datasets import FaultDiagnosisSmallTEP
from ice.fault_diagnosis.models import MLP

/Users/vitalijpozdnakov/miniconda3/envs/ice/lib/python3.9/site-packages/tqdm/auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
  from .autonotebook import tqdm as notebook_tqdm

dataset = FaultDiagnosisSmallTEP()

Downloading small_tep: 100%|██████████| 18.2M/18.2M [00:01<00:00, 10.1MB/s]
Extracting df.csv: 58.6MB [00:00, 153MB/s]                            
Extracting train_mask.csv: 9.77MB [00:00, 1.60GB/s]                   
Extracting target.csv: 9.77MB [00:00, 2.33GB/s]                   
Reading data/small_tep/df.csv: 100%|██████████| 153300/153300 [00:01<00:00, 79692.16it/s]
Reading data/small_tep/target.csv: 100%|██████████| 153300/153300 [00:00<00:00, 1771006.14it/s]
Reading data/small_tep/train_mask.csv: 100%|██████████| 153300/153300 [00:00<00:00, 1837204.89it/s]

dataset.df

		xmeas_1	xmeas_2	xmeas_3	xmeas_4	xmeas_5	xmeas_6	xmeas_7	xmeas_8	xmeas_9	xmeas_10	...	xmv_2	xmv_3	xmv_4	xmv_5	xmv_6	xmv_7	xmv_8	xmv_9	xmv_10	xmv_11
run_id	sample
413402073	1	0.25038	3674.0	4529.0	9.2320	26.889	42.402	2704.3	74.863	120.41	0.33818	...	53.744	24.657	62.544	22.137	39.935	42.323	47.757	47.510	41.258	18.447
	2	0.25109	3659.4	4556.6	9.4264	26.721	42.576	2705.0	75.000	120.41	0.33620	...	53.414	24.588	59.259	22.084	40.176	38.554	43.692	47.427	41.359	17.194
	3	0.25038	3660.3	4477.8	9.4426	26.875	42.070	2706.2	74.771	120.42	0.33563	...	54.357	24.666	61.275	22.380	40.244	38.990	46.699	47.468	41.199	20.530
	4	0.24977	3661.3	4512.1	9.4776	26.758	42.063	2707.2	75.224	120.39	0.33553	...	53.946	24.725	59.856	22.277	40.257	38.072	47.541	47.658	41.643	18.089
	5	0.29405	3679.0	4497.0	9.3381	26.889	42.650	2705.1	75.388	120.39	0.32632	...	53.658	28.797	60.717	21.947	39.144	41.955	47.645	47.346	41.507	18.461
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
312148819	956	0.24842	3694.2	4491.2	9.3946	26.780	42.655	2708.3	74.765	120.41	0.32959	...	53.891	24.580	63.320	21.867	38.868	36.061	48.088	45.470	41.463	17.078
	957	0.22612	3736.4	4523.1	9.3655	26.778	42.730	2711.0	75.142	120.38	0.32645	...	53.675	21.831	64.142	22.027	38.842	39.144	44.560	45.598	41.591	16.720
	958	0.22386	3692.8	4476.5	9.3984	26.673	42.528	2712.7	74.679	120.43	0.32484	...	54.233	22.053	59.228	22.235	39.040	35.116	45.737	45.490	41.884	16.310
	959	0.22561	3664.2	4483.0	9.4293	26.435	42.469	2710.2	74.857	120.38	0.31932	...	53.335	22.248	60.567	21.820	37.979	33.394	48.503	45.512	40.630	20.996
	960	0.22585	3717.6	4492.8	9.4061	26.869	42.176	2710.5	74.722	120.41	0.31926	...	53.217	22.225	63.429	22.259	37.986	34.810	47.810	45.639	41.898	18.378

153300 rows × 52 columns

dataset.target

run_id     sample
413402073  1          0
           2          0
           3          0
           4          0
           5          0
                     ..
312148819  956       20
           957       20
           958       20
           959       20
           960       20
Name: target, Length: 153300, dtype: int64

scaler = StandardScaler()
dataset.df[dataset.train_mask] = scaler.fit_transform(dataset.df[dataset.train_mask])
dataset.df[dataset.test_mask] = scaler.transform(dataset.df[dataset.test_mask])

model = MLP(window_size=10, lr=0.001, verbose=True)
model.fit(dataset.df[dataset.train_mask], dataset.target[dataset.train_mask])

Creating sequence of samples: 100%|██████████| 105/105 [00:00<00:00, 777.58it/s]
Epochs ...:  10%|█         | 1/10 [00:02<00:24,  2.70s/it]

Epoch 1, Loss: 0.4350

Epochs ...:  20%|██        | 2/10 [00:05<00:23,  2.88s/it]

Epoch 2, Loss: 0.5717

Epochs ...:  30%|███       | 3/10 [00:08<00:19,  2.82s/it]

Epoch 3, Loss: 0.3677

Epochs ...:  40%|████      | 4/10 [00:11<00:16,  2.79s/it]

Epoch 4, Loss: 0.6005

Epochs ...:  50%|█████     | 5/10 [00:13<00:13,  2.77s/it]

Epoch 5, Loss: 0.7325

Epochs ...:  60%|██████    | 6/10 [00:16<00:11,  2.86s/it]

Epoch 6, Loss: 0.3584

Epochs ...:  70%|███████   | 7/10 [00:19<00:08,  2.87s/it]

Epoch 7, Loss: 0.5394

Epochs ...:  80%|████████  | 8/10 [00:22<00:05,  2.81s/it]

Epoch 8, Loss: 0.3939

Epochs ...:  90%|█████████ | 9/10 [00:27<00:03,  3.57s/it]

Epoch 9, Loss: 0.3941

Epochs ...: 100%|██████████| 10/10 [00:35<00:00,  3.52s/it]

Epoch 10, Loss: 0.4601

metrics = model.evaluate(dataset.df[dataset.test_mask], dataset.target[dataset.test_mask])
metrics

Creating sequence of samples: 100%|██████████| 105/105 [00:00<00:00, 345.39it/s]
                                                             

{'accuracy': 0.7945664160401003}

Tutorial on fault diagnosis task#