Using evolution variables

Intermediate Predictive Analytics in Python

Nele Verbiest

Senior Data Scientist @PythonPredictions

Building predictive models

# Import the linear_model module
from sklearn import linear_model

# Predictive variables
variables = ["gender","age", "donations_last_year", "ratio_month_year"]

# Select predictors and target
X = basetable[variables]
y = basetable[["target"]]

# Construct the logistic regression model
logreg = linear_model.LogisticRegression()
logreg.fit(X, y)

Making predictions

# Import the linear_model module
from sklearn import linear_model

# Predictive variables
variables = ["gender","age", "donations_last_year", "ratio_month_year"]

# Select predictors and target
X = basetable[variables]
y = basetable[["target"]]

# Construct the logistic regression model
logreg = linear_model.LogisticRegression()
logreg.fit(X, y)

# Make predictions
predictions = logreg.predict_proba(X)[:,1]

Evaluating predictive models using AUC

# Import roc_auc_score module from sklearn.metrics
from sklearn.metrics import roc_auc_score

# Calculate the AUC
auc= roc_auc_score(y, predictions)
print(round(auc,2))

0.56

The predictor insight graph

# Discretize the variable in 5 bins and add to the basetable
basetable["ratio_month_year_disc"] = pd.qcut(basetable["ratio_month_year"], 5)

# Construct the predictor insight graph table
pig_table = create_pig_table(basetable, "target","ratio_month_year_disc")

```{python}
# Plot the predictor insight graph
plot_pig(pig_table, "ratio_month_year_disc")

Predictor insight graph interpretation

Let's practice!

Intermediate Predictive Analytics in Python