Holdout set¶

---

This example shows when and how to use ATOM's holdout set in an exploration pipeline.

The data used is a variation on the Australian weather dataset from Kaggle. You can download it from here. The goal of this dataset is to predict whether or not it will rain tomorrow training a binary classifier on target RainTomorrow.

Load the data¶

In [1]:

# Import packages
import pandas as pd
from atom import ATOMClassifier

# Import packages import pandas as pd from atom import ATOMClassifier

In [2]:

# Load data
X = pd.read_csv("./datasets/weatherAUS.csv")

# Let's have a look
X.head()

# Load data X = pd.read_csv("./datasets/weatherAUS.csv") # Let's have a look X.head()

Out[2]:

	Location	MinTemp	MaxTemp	Rainfall	Evaporation	Sunshine	WindGustDir	WindGustSpeed	WindDir9am	WindDir3pm	...	Humidity9am	Humidity3pm	Pressure9am	Pressure3pm	Cloud9am	Cloud3pm	Temp9am	Temp3pm	RainToday	RainTomorrow
0	MelbourneAirport	18.0	26.9	21.4	7.0	8.9	SSE	41.0	W	SSE	...	95.0	54.0	1019.5	1017.0	8.0	5.0	18.5	26.0	Yes	0
1	Adelaide	17.2	23.4	0.0	NaN	NaN	S	41.0	S	WSW	...	59.0	36.0	1015.7	1015.7	NaN	NaN	17.7	21.9	No	0
2	Cairns	18.6	24.6	7.4	3.0	6.1	SSE	54.0	SSE	SE	...	78.0	57.0	1018.7	1016.6	3.0	3.0	20.8	24.1	Yes	0
3	Portland	13.6	16.8	4.2	1.2	0.0	ESE	39.0	ESE	ESE	...	76.0	74.0	1021.4	1020.5	7.0	8.0	15.6	16.0	Yes	1
4	Walpole	16.4	19.9	0.0	NaN	NaN	SE	44.0	SE	SE	...	78.0	70.0	1019.4	1018.9	NaN	NaN	17.4	18.1	No	0

5 rows × 22 columns

Run the pipeline¶

In [3]:

# Initialize atom specifying a fraction of the dataset for holdout
atom = ATOMClassifier(X, n_rows=0.5, holdout_size=0.2, verbose=2, warnings=False)

# Initialize atom specifying a fraction of the dataset for holdout atom = ATOMClassifier(X, n_rows=0.5, holdout_size=0.2, verbose=2, warnings=False)

<< ================== ATOM ================== >>
Algorithm task: binary classification.

Dataset stats ==================== >>
Shape: (56877, 22)
Memory: 24.67 MB
Scaled: False
Missing values: 126549 (10.1%)
Categorical features: 5 (23.8%)
Duplicate samples: 16 (0.0%)
-------------------------------------
Train set size: 42658
Test set size: 14219
Holdout set size: 14219
-------------------------------------
|   |       dataset |         train |          test |
| - | ------------- | ------------- | ------------- |
| 0 |   44113 (3.5) |   33085 (3.5) |   11028 (3.5) |
| 1 |   12764 (1.0) |    9573 (1.0) |    3191 (1.0) |

In [4]:

# The test and holdout fractions are split after subsampling the dataset
# Also note that the holdout data set is not a part of atom's dataset
print("Length loaded data:", len(X))
print("Length dataset + holdout:", len(atom.dataset) + len(atom.holdout))

# The test and holdout fractions are split after subsampling the dataset # Also note that the holdout data set is not a part of atom's dataset print("Length loaded data:", len(X)) print("Length dataset + holdout:", len(atom.dataset) + len(atom.holdout))

Length loaded data: 142193
Length dataset + holdout: 71096

In [5]:

atom.impute()
atom.encode()

atom.impute() atom.encode()

Fitting Imputer...
Imputing missing values...
 --> Dropping 255 samples due to missing values in feature MinTemp.
 --> Dropping 124 samples due to missing values in feature MaxTemp.
 --> Dropping 607 samples due to missing values in feature Rainfall.
 --> Dropping 24204 samples due to missing values in feature Evaporation.
 --> Dropping 27058 samples due to missing values in feature Sunshine.
 --> Dropping 3828 samples due to missing values in feature WindGustDir.
 --> Dropping 3801 samples due to missing values in feature WindGustSpeed.
 --> Dropping 4078 samples due to missing values in feature WindDir9am.
 --> Dropping 1562 samples due to missing values in feature WindDir3pm.
 --> Dropping 553 samples due to missing values in feature WindSpeed9am.
 --> Dropping 1093 samples due to missing values in feature WindSpeed3pm.
 --> Dropping 705 samples due to missing values in feature Humidity9am.
 --> Dropping 1455 samples due to missing values in feature Humidity3pm.
 --> Dropping 5568 samples due to missing values in feature Pressure9am.
 --> Dropping 5565 samples due to missing values in feature Pressure3pm.
 --> Dropping 21344 samples due to missing values in feature Cloud9am.
 --> Dropping 22688 samples due to missing values in feature Cloud3pm.
 --> Dropping 359 samples due to missing values in feature Temp9am.
 --> Dropping 1095 samples due to missing values in feature Temp3pm.
 --> Dropping 607 samples due to missing values in feature RainToday.
Fitting Encoder...
Encoding categorical columns...
 --> LeaveOneOut-encoding feature Location. Contains 26 classes.
 --> LeaveOneOut-encoding feature WindGustDir. Contains 16 classes.
 --> LeaveOneOut-encoding feature WindDir9am. Contains 16 classes.
 --> LeaveOneOut-encoding feature WindDir3pm. Contains 16 classes.
 --> Ordinal-encoding feature RainToday. Contains 2 classes.

In [6]:

# Unlike train and test, the holdout data set is not transformed until used for predictions
atom.holdout

# Unlike train and test, the holdout data set is not transformed until used for predictions atom.holdout

Out[6]:

	Location	MinTemp	MaxTemp	Rainfall	Evaporation	Sunshine	WindGustDir	WindGustSpeed	WindDir9am	WindDir3pm	...	Humidity9am	Humidity3pm	Pressure9am	Pressure3pm	Cloud9am	Cloud3pm	Temp9am	Temp3pm	RainToday	RainTomorrow
0	Albany	10.8	19.8	0.4	1.4	9.4	NaN	NaN	N	NNW	...	79.0	53.0	1019.0	1014.9	2.0	2.0	14.2	19.5	No	0
1	NorahHead	14.2	21.3	52.0	NaN	NaN	SSE	56.0	S	SSE	...	69.0	70.0	1015.5	1013.6	NaN	NaN	18.5	19.6	Yes	1
2	SalmonGums	10.7	16.5	0.4	NaN	NaN	WNW	72.0	WNW	WNW	...	77.0	61.0	NaN	NaN	NaN	NaN	12.3	11.9	No	0
3	Tuggeranong	11.4	26.9	12.8	NaN	NaN	ENE	37.0	WSW	NE	...	85.0	54.0	1016.8	1014.6	NaN	NaN	15.5	26.0	Yes	1
4	Richmond	8.5	18.2	0.0	NaN	NaN	WSW	31.0	NNE	E	...	98.0	55.0	1027.4	1024.3	8.0	8.0	10.8	16.6	No	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
14214	Nuriootpa	10.1	20.1	0.0	4.0	7.0	SE	35.0	E	ESE	...	49.0	35.0	1027.7	1024.0	2.0	NaN	14.7	19.0	No	0
14215	Sale	9.3	17.5	0.0	4.4	4.9	W	78.0	NW	WNW	...	52.0	46.0	1010.7	1011.2	6.0	5.0	14.1	15.9	No	0
14216	Bendigo	8.3	25.1	0.0	NaN	NaN	SSE	35.0	SSE	SSW	...	51.0	23.0	1017.7	1014.8	NaN	NaN	13.1	23.6	No	0
14217	Tuggeranong	9.0	27.0	0.0	NaN	NaN	N	31.0	NW	NNW	...	74.0	30.0	1017.9	1013.5	NaN	NaN	14.7	25.3	No	0
14218	BadgerysCreek	13.2	26.5	0.0	NaN	NaN	WSW	37.0	WSW	SW	...	49.0	28.0	1014.2	1013.8	NaN	NaN	19.8	25.6	No	0

14219 rows × 22 columns

In [7]:

atom.run(models=["GNB", "LR", "RF"])

atom.run(models=["GNB", "LR", "RF"])

Training ========================= >>
Models: GNB, LR, RF
Metric: f1


Results for Gaussian Naive Bayes:
Fit ---------------------------------------------
Train evaluation --> f1: 0.6126
Test evaluation --> f1: 0.5915
Time elapsed: 0.035s
-------------------------------------------------
Total time: 0.035s


Results for Logistic Regression:
Fit ---------------------------------------------
Train evaluation --> f1: 0.6325
Test evaluation --> f1: 0.6014
Time elapsed: 0.131s
-------------------------------------------------
Total time: 0.131s


Results for Random Forest:
Fit ---------------------------------------------
Train evaluation --> f1: 1.0
Test evaluation --> f1: 0.6149
Time elapsed: 2.839s
-------------------------------------------------
Total time: 2.839s


Final results ==================== >>
Duration: 3.005s
-------------------------------------
Gaussian Naive Bayes --> f1: 0.5915
Logistic Regression  --> f1: 0.6014
Random Forest        --> f1: 0.6149 ~ !

In [8]:

atom.plot_prc()

atom.plot_prc()

In [9]:

# Based on the results on the test set, we select the best model for further tuning
atom.run("lr_tuned", n_calls=10, n_initial_points=5)

# Based on the results on the test set, we select the best model for further tuning atom.run("lr_tuned", n_calls=10, n_initial_points=5)

Training ========================= >>
Models: LR_tuned
Metric: f1


Running BO for Logistic Regression...
| call             | penalty |       C |  solver | max_iter | l1_ratio |      f1 | best_f1 |    time | total_time |
| ---------------- | ------- | ------- | ------- | -------- | -------- | ------- | ------- | ------- | ---------- |
| Initial point 1  |      l2 |   0.063 | libli.. |      498 |      --- |  0.6373 |  0.6373 |  0.633s |     0.646s |
| Initial point 2  |    none |     --- |    saga |      625 |      --- |  0.6142 |  0.6373 |  0.746s |     1.628s |
| Initial point 3  |      l2 |  0.6394 | newto.. |      174 |      --- |  0.6226 |  0.6373 |  0.568s |     2.269s |
| Initial point 4  |      l2 |  0.1383 | libli.. |      659 |      --- |  0.6191 |  0.6373 |  0.530s |     2.874s |
| Initial point 5  |      l2 | 11.7507 |     sag |      603 |      --- |  0.6343 |  0.6373 |  0.589s |     3.536s |
| Iteration 6      |      l2 |  0.9379 |   lbfgs |      575 |      --- |  0.6218 |  0.6373 |  0.542s |     4.372s |
| Iteration 7      |      l1 |  0.0276 | libli.. |      375 |      --- |  0.6337 |  0.6373 |  0.566s |     5.224s |
| Iteration 8      |      l1 |  7.9949 |    saga |      856 |      --- |  0.6306 |  0.6373 |  0.721s |     6.277s |
| Iteration 9      |      l2 |  0.0011 |   lbfgs |      179 |      --- |  0.5654 |  0.6373 |  0.571s |     7.152s |
| Iteration 10     |      l2 |   100.0 |   lbfgs |      100 |      --- |   0.633 |  0.6373 |  0.542s |     8.097s |
Bayesian Optimization ---------------------------
Best call --> Initial point 1
Best parameters --> {'penalty': 'l2', 'C': 0.063, 'solver': 'liblinear', 'max_iter': 498}
Best evaluation --> f1: 0.6373
Time elapsed: 8.612s
Fit ---------------------------------------------
Train evaluation --> f1: 0.6348
Test evaluation --> f1: 0.5979
Time elapsed: 0.126s
-------------------------------------------------
Total time: 8.739s


Final results ==================== >>
Duration: 8.739s
-------------------------------------
Logistic Regression --> f1: 0.5979

Analyze the results¶

We already used the test set to choose the best model for futher tuning, so this set is no longer truly independent. Although it may not be directly visible in the results, using the test set now to evaluate the tuned LR model would be a mistake, since it carries a bias. For this reason, we have set apart an extra, indepedent set to validate the final model: the holdout set. If we are not going to use the test set for validation, we might as well use it to train the model and so optimize the use of the available data. Use the full_train method for this.

In [10]:

# Re-train the model on the full dataset (train + test) 
atom.lr_tuned.full_train()

# Re-train the model on the full dataset (train + test) atom.lr_tuned.full_train()

Model LR_tuned successfully retrained.

In [11]:

# Evaluate on the holdout set
atom.lr_tuned.evaluate(dataset="holdout")

# Evaluate on the holdout set atom.lr_tuned.evaluate(dataset="holdout")

Out[11]:

accuracy             0.848277
average_precision    0.711065
balanced_accuracy    0.730813
f1                   0.604979
jaccard              0.433670
matthews_corrcoef    0.525790
precision            0.727684
recall               0.517685
roc_auc              0.882586
Name: LR_tuned, dtype: float64

In [12]:

atom.lr_tuned.plot_prc(dataset="holdout")

atom.lr_tuned.plot_prc(dataset="holdout")