Using Text Data with EvalML

In this demo, we will show you how to use EvalML to build models which use text data.

import evalml
from evalml import AutoMLSearch

Dataset

We will be utilizing a dataset of SMS text messages, some of which are categorized as spam, and others which are not (“ham”). This dataset is originally from Kaggle, but modified to produce a slightly more even distribution of spam to ham.

from urllib.request import urlopen
import pandas as pd

input_data = urlopen('https://featurelabs-static.s3.amazonaws.com/spam_text_messages_modified.csv')
data = pd.read_csv(input_data)

X = data.drop(['Category'], axis=1)
y = data['Category']

display(X.head())

	Message
0	Free entry in 2 a wkly comp to win FA Cup fina...
1	FreeMsg Hey there darling it's been 3 week's n...
2	WINNER!! As a valued network customer you have...
3	Had your mobile 11 months or more? U R entitle...
4	SIX chances to win CASH! From 100 to 20,000 po...

The ham vs spam distribution of the data is 3:1, so any machine learning model must get above 75% accuracy in order to perform better than a trivial baseline model which simply classifies everything as ham.

y.value_counts(normalize=True)

ham     0.750084
spam    0.249916
Name: Category, dtype: float64

Search for best pipeline

In order to validate the results of the pipeline creation and optimization process, we will save some of our data as a holdout set.

X_train, X_holdout, y_train, y_holdout = evalml.preprocessing.split_data(X, y, problem_type='binary', test_size=0.2, random_seed=0)

EvalML uses Woodwork to automatically detect which columns are text columns, so you can run search normally, as you would if there was no text data. We can print out the logical type of the Message column and assert that it is indeed inferred as a natural language column.

X_train.types

	Physical Type	Logical Type	Semantic Tag(s)
Data Column
Message	string	NaturalLanguage	[]

Because the spam/ham labels are binary, we will use AutoMLSearch(X_train=X_train, y_train=y_train, problem_type='binary'). When we call .search(), the search for the best pipeline will begin.

automl = AutoMLSearch(X_train=X_train, y_train=y_train,
                      problem_type='binary',
                      max_batches=1,
                      optimize_thresholds=True)

automl.search()

Generating pipelines to search over...
*****************************
* Beginning pipeline search *
*****************************

Optimizing for Log Loss Binary.
Lower score is better.

Using SequentialEngine to train and score pipelines.
Searching up to 1 batches for a total of 9 pipelines.
Allowed model families: xgboost, lightgbm, decision_tree, catboost, extra_trees, random_forest, linear_model

Batch 1: (1/9) Mode Baseline Binary Classification P... Elapsed:00:00
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 8.638
Batch 1: (2/9) Decision Tree Classifier w/ Text Feat... Elapsed:00:00
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.802
High coefficient of variation (cv >= 0.2) within cross validation scores. Decision Tree Classifier w/ Text Featurization Component may not perform as estimated on unseen data.
Batch 1: (3/9) LightGBM Classifier w/ Text Featuriza... Elapsed:00:10
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.215
High coefficient of variation (cv >= 0.2) within cross validation scores. LightGBM Classifier w/ Text Featurization Component may not perform as estimated on unseen data.
Batch 1: (4/9) Extra Trees Classifier w/ Text Featur... Elapsed:00:19
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.252
High coefficient of variation (cv >= 0.2) within cross validation scores. Extra Trees Classifier w/ Text Featurization Component may not perform as estimated on unseen data.
Batch 1: (5/9) Elastic Net Classifier w/ Text Featur... Elapsed:00:28
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.543
Batch 1: (6/9) CatBoost Classifier w/ Text Featuriza... Elapsed:00:37
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.526
Batch 1: (7/9) XGBoost Classifier w/ Text Featurizat... Elapsed:00:46
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.179
High coefficient of variation (cv >= 0.2) within cross validation scores. XGBoost Classifier w/ Text Featurization Component may not perform as estimated on unseen data.
Batch 1: (8/9) Random Forest Classifier w/ Text Feat... Elapsed:00:55
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.155
High coefficient of variation (cv >= 0.2) within cross validation scores. Random Forest Classifier w/ Text Featurization Component may not perform as estimated on unseen data.
Batch 1: (9/9) Logistic Regression Classifier w/ Tex... Elapsed:01:05
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.214
High coefficient of variation (cv >= 0.2) within cross validation scores. Logistic Regression Classifier w/ Text Featurization Component + Standard Scaler may not perform as estimated on unseen data.

Search finished after 01:15
Best pipeline: Random Forest Classifier w/ Text Featurization Component
Best pipeline Log Loss Binary: 0.154849

View rankings and select pipeline

Once the fitting process is done, we can see all of the pipelines that were searched.

automl.rankings

	id	pipeline_name	score	validation_score	percent_better_than_baseline	high_variance_cv	parameters
0	7	Random Forest Classifier w/ Text Featurization...	0.154849	0.110302	98.207418	True	{'Random Forest Classifier': {'n_estimators': ...
1	6	XGBoost Classifier w/ Text Featurization Compo...	0.178639	0.113254	97.932010	True	{'XGBoost Classifier': {'eta': 0.1, 'max_depth...
2	8	Logistic Regression Classifier w/ Text Featuri...	0.214011	0.165624	97.522538	True	{'Logistic Regression Classifier': {'penalty':...
3	2	LightGBM Classifier w/ Text Featurization Comp...	0.214580	0.136260	97.515944	True	{'LightGBM Classifier': {'boosting_type': 'gbd...
4	3	Extra Trees Classifier w/ Text Featurization C...	0.252206	0.216198	97.080377	True	{'Extra Trees Classifier': {'n_estimators': 10...
5	5	CatBoost Classifier w/ Text Featurization Comp...	0.526403	0.512717	93.906174	False	{'CatBoost Classifier': {'n_estimators': 10, '...
6	4	Elastic Net Classifier w/ Text Featurization C...	0.542803	0.529152	93.716325	False	{'Elastic Net Classifier': {'alpha': 0.5, 'l1_...
7	1	Decision Tree Classifier w/ Text Featurization...	0.801766	0.555179	90.718481	True	{'Decision Tree Classifier': {'criterion': 'gi...
8	0	Mode Baseline Binary Classification Pipeline	8.638305	8.623860	0.000000	False	{'Baseline Classifier': {'strategy': 'mode'}}

To select the best pipeline we can call automl.best_pipeline.

best_pipeline = automl.best_pipeline

Describe pipeline

You can get more details about any pipeline, including how it performed on other objective functions.

automl.describe_pipeline(automl.rankings.iloc[0]["id"])

************************************************************
* Random Forest Classifier w/ Text Featurization Component *
************************************************************

Problem Type: binary
Model Family: Random Forest

Pipeline Steps
==============
1. Text Featurization Component
2. Random Forest Classifier
         * n_estimators : 100
         * max_depth : 6
         * n_jobs : -1

Training
========
Training for binary problems.
Total training time (including CV): 9.5 seconds

Cross Validation
----------------
             Log Loss Binary  MCC Binary   AUC  Precision    F1  Balanced Accuracy Binary  Accuracy Binary # Training # Validation
0                      0.110       0.895 0.987      0.938 0.921                     0.942            0.961     1594.0        797.0
1                      0.144       0.854 0.980      0.919 0.888                     0.917            0.946     1594.0        797.0
2                      0.210       0.783 0.962      0.839 0.837                     0.891            0.918     1594.0        797.0
mean                   0.155       0.844 0.977      0.899 0.882                     0.917            0.942          -            -
std                    0.051       0.057 0.013      0.052 0.042                     0.026            0.022          -            -
coef of var            0.326       0.067 0.013      0.058 0.048                     0.028            0.023          -            -

best_pipeline.graph()

Notice above that there is a Text Featurization Component as the second step in the pipeline. The Woodwork DataTable passed in to AutoML search recognizes that 'Message' is a text column, and converts this text into numerical values that can be handled by the estimator.

Evaluate on holdout

Now, we can score the pipeline on the holdout data using the core objectives for binary classification problems.

scores = best_pipeline.score(X_holdout, y_holdout,  objectives=evalml.objectives.get_core_objectives('binary'))
print(f'Accuracy Binary: {scores["Accuracy Binary"]}')

Accuracy Binary: 0.9732441471571907

As you can see, this model performs relatively well on this dataset, even on unseen data.

Why encode text this way?

To demonstrate the importance of text-specific modeling, let’s train a model with the same dataset, without letting AutoMLSearch detect the text column. We can change this by explicitly setting the data type of the 'Message' column in Woodwork to Categorical using the utility method infer_feature_types.

from evalml.utils import infer_feature_types
X = infer_feature_types(X, {'Message': 'Categorical'})
X_train, X_holdout, y_train, y_holdout = evalml.preprocessing.split_data(X, y, problem_type='binary', test_size=0.2, random_seed=0)

automl_no_text = AutoMLSearch(X_train=X_train, y_train=y_train,
                              problem_type='binary',
                              max_batches=1,
                              optimize_thresholds=True)

automl_no_text.search()

Generating pipelines to search over...
*****************************
* Beginning pipeline search *
*****************************

Optimizing for Log Loss Binary.
Lower score is better.

Using SequentialEngine to train and score pipelines.
Searching up to 1 batches for a total of 9 pipelines.
Allowed model families: xgboost, lightgbm, decision_tree, catboost, extra_trees, random_forest, linear_model

Batch 1: (1/9) Mode Baseline Binary Classification P... Elapsed:00:00
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 8.638
Batch 1: (2/9) Decision Tree Classifier w/ Imputer +... Elapsed:00:00
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.561
Batch 1: (3/9) LightGBM Classifier w/ Imputer + One ... Elapsed:00:00
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.562
Batch 1: (4/9) Extra Trees Classifier w/ Imputer + O... Elapsed:00:00
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.561
Batch 1: (5/9) Elastic Net Classifier w/ Imputer + O... Elapsed:00:01
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.563
Batch 1: (6/9) CatBoost Classifier w/ Imputer           Elapsed:00:02
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.622
Batch 1: (7/9) XGBoost Classifier w/ Imputer + One H... Elapsed:00:02
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.562
Batch 1: (8/9) Random Forest Classifier w/ Imputer +... Elapsed:00:02
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.561
Batch 1: (9/9) Logistic Regression Classifier w/ Imp... Elapsed:00:03
        Starting cross validation
        Finished cross validation - mean Log Loss Binary: 0.561

Search finished after 00:04
Best pipeline: Logistic Regression Classifier w/ Imputer + One Hot Encoder + Standard Scaler
Best pipeline Log Loss Binary: 0.560554

Like before, we can look at the rankings and pick the best pipeline.

automl_no_text.rankings

	id	pipeline_name	score	validation_score	percent_better_than_baseline	high_variance_cv	parameters
0	8	Logistic Regression Classifier w/ Imputer + On...	0.560554	0.558141	93.510838	False	{'Imputer': {'categorical_impute_strategy': 'm...
1	1	Decision Tree Classifier w/ Imputer + One Hot ...	0.561003	0.558148	93.505636	False	{'Imputer': {'categorical_impute_strategy': 'm...
2	7	Random Forest Classifier w/ Imputer + One Hot ...	0.561179	0.559141	93.503593	False	{'Imputer': {'categorical_impute_strategy': 'm...
3	3	Extra Trees Classifier w/ Imputer + One Hot En...	0.561247	0.559029	93.502811	False	{'Imputer': {'categorical_impute_strategy': 'm...
4	6	XGBoost Classifier w/ Imputer + One Hot Encoder	0.562197	0.561991	93.491811	False	{'Imputer': {'categorical_impute_strategy': 'm...
5	2	LightGBM Classifier w/ Imputer + One Hot Encoder	0.562451	0.561991	93.488872	False	{'Imputer': {'categorical_impute_strategy': 'm...
6	4	Elastic Net Classifier w/ Imputer + One Hot En...	0.562556	0.562070	93.487658	False	{'Imputer': {'categorical_impute_strategy': 'm...
7	5	CatBoost Classifier w/ Imputer	0.622449	0.622812	92.794318	False	{'Imputer': {'categorical_impute_strategy': 'm...
8	0	Mode Baseline Binary Classification Pipeline	8.638305	8.623860	0.000000	False	{'Baseline Classifier': {'strategy': 'mode'}}

best_pipeline_no_text = automl_no_text.best_pipeline

Here, changing the data type of the text column removed the Text Featurization Component from the pipeline.

best_pipeline_no_text.graph()

automl_no_text.describe_pipeline(automl_no_text.rankings.iloc[0]["id"])

*********************************************************************************
* Logistic Regression Classifier w/ Imputer + One Hot Encoder + Standard Scaler *
*********************************************************************************

Problem Type: binary
Model Family: Linear

Pipeline Steps
==============
1. Imputer
         * categorical_impute_strategy : most_frequent
         * numeric_impute_strategy : mean
         * categorical_fill_value : None
         * numeric_fill_value : None
2. One Hot Encoder
         * top_n : 10
         * features_to_encode : None
         * categories : None
         * drop : if_binary
         * handle_unknown : ignore
         * handle_missing : error
3. Standard Scaler
4. Logistic Regression Classifier
         * penalty : l2
         * C : 1.0
         * n_jobs : -1
         * multi_class : auto
         * solver : lbfgs

Training
========
Training for binary problems.
Total training time (including CV): 0.4 seconds

Cross Validation
----------------
             Log Loss Binary  MCC Binary   AUC  Precision    F1  Balanced Accuracy Binary  Accuracy Binary # Training # Validation
0                      0.558       0.061 0.508      1.000 0.010                     0.503            0.752     1594.0        797.0
1                      0.561       0.000 0.503      0.000 0.000                     0.500            0.750     1594.0        797.0
2                      0.563       0.000 0.503      0.000 0.000                     0.500            0.749     1594.0        797.0
mean                   0.561       0.020 0.504      0.333 0.003                     0.501            0.750          -            -
std                    0.002       0.035 0.003      0.577 0.006                     0.001            0.001          -            -
coef of var            0.004       1.732 0.006      1.732 1.732                     0.003            0.002          -            -

# get standard performance metrics on holdout data
scores = best_pipeline_no_text.score(X_holdout, y_holdout, objectives=evalml.objectives.get_core_objectives('binary'))
print(f'Accuracy Binary: {scores["Accuracy Binary"]}')

Accuracy Binary: 0.7525083612040134

Without the Text Featurization Component, the 'Message' column was treated as a categorical column, and therefore the conversion of this text to numerical features happened in the One Hot Encoder. The best pipeline encoded the top 10 most frequent “categories” of these texts, meaning 10 text messages were one-hot encoded and all the others were dropped. Clearly, this removed almost all of the information from the dataset, as we can see the best_pipeline_no_text did not beat the random guess of predicting “ham” in every case.

Using the Cost-Benefit Matrix Objective User Guide