import pandas as pd
from sklearn.linear_model import LogisticRegression
from utils import impute_nas, send_submission
train = pd.read_csv("input/dev.csv") #DEV-SAMPLE
test = pd.read_csv("input/oot0.csv") #OUT-OF-TIME SAMPLE
TARGET = 'ob_target'
model = LogisticRegression()
model.fit(train.drop(TARGET, axis=1), train[TARGET])
preds = model.predict_proba(impute_nas(test))[:, 1]
send_submission("logit_all_vars.csv", preds)
'max_depth': [10, 20, 30, 40, 50, 60, 70, 80, 90, 100, None],
'max_features': ['log2', 'sqrt', None],
'min_samples_leaf': [1, 2, 4],
'min_samples_split': [2, 5, 10]
}
create_file = True
for sampling in params["bootstrap"]:
for depth in params["max_depth"]:
for max_feat in params["max_features"]:
for min_leaf in params["min_samples_leaf"]:
for min_split in params["min_samples_split"]:
p = {
'bootstrap': sampling,
'max_depth': depth,
'max_features': max_feat,
'min_samples_leaf': min_leaf,
'min_samples_split': min_split,
"random_state": SEED,
"n_estimators": 1000
}
model = RandomForestClassifier(**p).fit(X, y)
preds = model.predict_proba(test)[:, 1]
results = send_submission("doesnt_matter.csv", preds)
save_hyperparameters(p, results, "grid_search_rf.csv",
create_file)
create_file = False
sleep(3)
import xgboost as xgb
import pandas as pd
from utils import impute_nas, send_submission
train = pd.read_csv("input/dev.csv") #DEV-SAMPLE
test = pd.read_csv("input/oot0.csv") #OUT-OF-TIME SAMPLE
TARGET = 'ob_target'
model = xgb.XGBClassifier(random_state=20190626, n_estimators=1000)
model.fit(train.drop(TARGET, axis=1), train[TARGET])
preds = model.predict_proba(impute_nas(test))[:, 1]
send_submission("xgb_all_vars.csv", preds)
from sklearn.ensemble import RandomForestClassifier
import seaborn as sns
import matplotlib.pyplot as plt
train = pd.read_csv("input/dev.csv").drop("id", axis=1) #DEV-SAMPLE
test = impute_nas(pd.read_csv("input/oot0.csv").drop(
"id", axis=1)) #OUT-OF-TIME SAMPLE
model = RandomForestClassifier(random_state=20190628, n_estimators=1000)
results = perform_rfe(model,
train,
test,
"random_forest_rfe.csv",
to_remove=80)
#results = pd.read_csv("random_forest_rfe.csv")
num_removed = 82 - results["n"]
fig, ax = plt.subplots(figsize=(8, 8))
sns.scatterplot(x=num_removed[:70], y=results["grade"][:70], ax=ax)
best = results[results["grade"] == max(results["grade"])]
model = RandomForestClassifier(random_state=20190628, n_estimators=1000)
model.fit(train[best["feats"].item().split(";")], train[TARGET])
preds = model.predict_proba(test[best["feats"].item().split(";")])[:, 1]
result = send_submission("rf_after_rfe.csv", preds)