1. Select a dataset with binary target values using https://machinelearningmastery.com/standard-machine-learning-datasets/
e.g. banknote or diabetes dataset
2. Use pandas to read CSV file as dataframe. e.g. The following code helps import pima diabetes dataset
col_names = ['pregnant', 'glucose', 'bp', 'skin', 'insulin', 'bmi', 'pedigree', 'age', 'label']
# load dataset
pima = pd.read_csv("pima-indians-diabetes-database.csv", header=None, names=col_names)
3. Select 5 (if not possible then select 4) features from the chosen dataset. List all features you selected in your report.
For example, the following code will select two features
feature_cols = ['pregnant', 'age']
X = pima[feature_cols]
4. Use “train _test_split” from “sklearn.cross_validationtrain” to split test and training data by 40% testing + 60% training.
5. Fit your model with training data and test your model after fitting.
6. Calculate and plot out
the confusion matrix
precision score, recall score, F score
Copy your console output (these scores) to your report.
7. Plot out the ROC curve and print out the ROC_AUC score (sklearn.metrics.roc_curve() and sklearn.metrics.roc_auc_score() can be used.)
