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)
Scaled: False
Missing values: 126517 (10.1%)
Categorical features: 5 (23.8%)
Duplicate samples: 19 (0.0%)
-------------------------------------
Train set size: 42658
Test set size: 14219
Holdout set size: 14219
-------------------------------------
|    |       dataset |         train |          test |
| -- | ------------- | ------------- | ------------- |
| 0  |   44192 (3.5) |   33244 (3.5) |   10948 (3.3) |
| 1  |   12685 (1.0) |    9414 (1.0) |    3271 (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 272 samples due to missing values in feature MinTemp.
 --> Dropping 137 samples due to missing values in feature MaxTemp.
 --> Dropping 590 samples due to missing values in feature Rainfall.
 --> Dropping 24144 samples due to missing values in feature Evaporation.
 --> Dropping 26865 samples due to missing values in feature Sunshine.
 --> Dropping 3800 samples due to missing values in feature WindGustDir.
 --> Dropping 3772 samples due to missing values in feature WindGustSpeed.
 --> Dropping 4070 samples due to missing values in feature WindDir9am.
 --> Dropping 1530 samples due to missing values in feature WindDir3pm.
 --> Dropping 569 samples due to missing values in feature WindSpeed9am.
 --> Dropping 1075 samples due to missing values in feature WindSpeed3pm.
 --> Dropping 722 samples due to missing values in feature Humidity9am.
 --> Dropping 1483 samples due to missing values in feature Humidity3pm.
 --> Dropping 5601 samples due to missing values in feature Pressure9am.
 --> Dropping 5586 samples due to missing values in feature Pressure3pm.
 --> Dropping 21440 samples due to missing values in feature Cloud9am.
 --> Dropping 22764 samples due to missing values in feature Cloud3pm.
 --> Dropping 375 samples due to missing values in feature Temp9am.
 --> Dropping 1132 samples due to missing values in feature Temp3pm.
 --> Dropping 590 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	Ballarat	8.3	31.5	0.0	NaN	NaN	NW	43.0	N	WNW	...	39.0	20.0	1014.9	1012.5	NaN	1.0	22.7	31.2	No	0
1	WaggaWagga	9.9	25.8	0.0	2.4	7.6	E	22.0	ENE	E	...	81.0	48.0	1016.0	1011.5	1.0	4.0	16.7	24.0	No	0
2	Ballarat	4.8	24.3	0.0	NaN	NaN	ESE	37.0	SSE	SSE	...	51.0	25.0	1020.2	1018.1	NaN	NaN	13.1	22.4	No	0
3	Williamtown	12.2	26.6	0.0	2.0	NaN	NW	19.0	NW	E	...	91.0	51.0	1025.6	1021.9	NaN	NaN	17.0	25.8	No	0
4	Newcastle	7.6	23.8	0.0	NaN	NaN	NaN	NaN	NaN	NaN	...	92.0	NaN	NaN	NaN	0.0	NaN	12.0	NaN	No	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
14214	Cobar	15.3	28.9	0.0	7.0	NaN	NaN	NaN	NE	SSE	...	42.0	18.0	1022.4	1019.4	1.0	2.0	21.9	27.8	No	0
14215	Witchcliffe	3.3	18.3	0.2	NaN	NaN	SSW	19.0	NaN	SE	...	98.0	59.0	1029.6	1028.0	NaN	NaN	9.4	17.0	No	0
14216	Nuriootpa	9.4	22.9	0.0	2.8	5.3	NNW	28.0	NE	NNW	...	44.0	23.0	1025.8	1023.7	6.0	6.0	14.3	22.0	No	0
14217	Sale	13.8	22.3	0.0	4.2	1.7	S	31.0	ESE	S	...	77.0	75.0	1010.6	1008.9	8.0	8.0	19.1	20.7	No	0
14218	Witchcliffe	9.7	26.9	0.0	NaN	NaN	N	44.0	E	SSE	...	68.0	69.0	1014.0	1008.5	NaN	NaN	15.7	20.0	No	1

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: 0.6027
Test evaluation --> f1: 0.6215
Time elapsed: 0.031s
-------------------------------------------------
Total time: 0.031s


Results for Logistic Regression:
Fit ---------------------------------------------
Train evaluation --> f1: 0.609
Test evaluation --> f1: 0.6471
Time elapsed: 0.125s
-------------------------------------------------
Total time: 0.125s


Results for Random Forest:
Fit ---------------------------------------------
Train evaluation --> f1: 0.9999
Test evaluation --> f1: 0.6465
Time elapsed: 2.969s
-------------------------------------------------
Total time: 2.969s


Final results ==================== >>
Duration: 3.125s
-------------------------------------
Gaussian Naive Bayes --> f1: 0.6215
Logistic Regression  --> f1: 0.6471 !
Random Forest        --> f1: 0.6465 ~

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 |  7.4773 | newto.. |      135 |      --- |  0.6182 |  0.6182 |  0.281s |     0.297s |
| Initial point 2  |      l2 |  0.0056 |     sag |      123 |      --- |  0.5994 |  0.6182 |  0.250s |     0.734s |
| Initial point 3  |      l2 |  0.0666 | newto.. |      573 |      --- |  0.6103 |  0.6182 |  0.266s |     1.063s |
| Initial point 4  |      l2 | 13.7037 | libli.. |      705 |      --- |  0.5997 |  0.6182 |  0.266s |     1.391s |
| Initial point 5  |      l2 |  0.0292 | libli.. |      550 |      --- |  0.6298 |  0.6298 |  0.250s |     1.703s |
| Iteration 6      |    none |     --- |   lbfgs |      144 |      --- |  0.6348 |  0.6348 |  0.250s |     2.219s |
| Iteration 7      |    none |     --- |     sag |      100 |      --- |  0.6209 |  0.6348 |  0.297s |     2.844s |
| Iteration 8      |      l2 |  0.2678 | newto.. |      703 |      --- |  0.6057 |  0.6348 |  0.281s |     3.407s |
| Iteration 9      |      l2 |  4.2629 | libli.. |      492 |      --- |  0.6455 |  0.6455 |  0.234s |     3.938s |
| Iteration 10     |      l2 |  3.5302 |     sag |      542 |      --- |  0.6559 |  0.6559 |  0.313s |     4.563s |
Bayesian Optimization ---------------------------
Best call --> Iteration 10
Best parameters --> {'penalty': 'l2', 'C': 3.5302, 'solver': 'sag', 'max_iter': 542}
Best evaluation --> f1: 0.6559
Time elapsed: 4.985s
Fit ---------------------------------------------
Train evaluation --> f1: 0.6094
Test evaluation --> f1: 0.6471
Time elapsed: 0.203s
-------------------------------------------------
Total time: 5.188s


Final results ==================== >>
Duration: 5.188s
-------------------------------------
Logistic Regression --> f1: 0.6471

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             0.854024
average_precision    0.727781
balanced_accuracy    0.742895
f1                   0.627017
jaccard              0.456682
matthews_corrcoef    0.550645
precision            0.749724
recall               0.538827
roc_auc              0.878702
Name: LR_tuned, dtype: float64

In [12]:

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

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