Created
June 19, 2017 02:37
-
-
Save ackjake/f4740dbd8b74d99348d952abed12ce1a to your computer and use it in GitHub Desktop.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
def get_target_mean(df_train, df_test, var, target, NFOLDS, NOISE): | |
""" | |
Creates out of fold averages by the categorical variable passed. | |
Decreasing the number of folds and increasing the noise can help | |
prevent over fitting the training set. | |
""" | |
df_train['mean_{}'.format(var)] = np.nan | |
df_test['mean_{}'.format(var)] = np.nan | |
values = np.empty((NFOLD)) | |
values[:] = np.nan | |
mean_stats = dict((el,np.copy(values)) for el in df_train[var].unique()) | |
kf = KFold(NFOLD, shuffle=True) | |
for i, (train_index, test_index) in enumerate(kf.split(df_train)): | |
cv_train = df_train.iloc[train_index] | |
cv_test = df_train.iloc[test_index] | |
grouped = cv_train.groupby(var)[[target]].mean().reset_index() | |
for v in grouped[var].unique(): | |
var_index = cv_test[cv_test[var] == v].index | |
impute_mean = grouped[grouped[var] == v][target].mean() | |
df_train.loc[var_index, 'mean_{}'.format(var)] = impute_mean * (np.random.uniform(0,0.01)+1) | |
mean_stats[v][i] = the_mean | |
avgDict = {} | |
for k,v in mean_stats.items(): | |
avgDict[k] = sum(v)/ float(len(v)) | |
for v in avgDict: | |
var_index = df_test[df_test[var] == v].index | |
df_test.loc[var_index, 'mean_{}'.format(var)] = avgDict[v] | |
df_train.drop(var, axis=1, inplace=True) | |
df_test.drop(var, axis=1, inplace=True) | |
return df_train, df_test |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment