From 811b7c6dcb5fd56fe07db36384a4d506dbbac061 Mon Sep 17 00:00:00 2001
From: NIPS-BRITS <emingcup2012@163.com>
Date: Wed, 28 Nov 2018 09:10:38 -0500
Subject: [PATCH] add rf baseline

---
 result/light_gbm.py     | 37 -------------------------------------
 result/random_forest.py | 27 +++++++++++++++++++++++++++
 2 files changed, 27 insertions(+), 37 deletions(-)
 delete mode 100644 result/light_gbm.py
 create mode 100644 result/random_forest.py

diff --git a/result/light_gbm.py b/result/light_gbm.py
deleted file mode 100644
index cdc0f8f7..00000000
--- a/result/light_gbm.py
+++ /dev/null
@@ -1,37 +0,0 @@
-import xgboost as xgb
-import numpy as np
-
-model_name = 'm_rnn'
-
-impute = np.load('./{}_data.npy'.format(model_name)).reshape(-1, 48 * 35)
-label = np.load('./{}_label.npy'.format(model_name))
-
-#data = fill
-data = impute
-
-n_train = 3000
-
-print(impute.shape)
-print(label.shape)
-
-dtrain = xgb.DMatrix(data[:n_train], label = label[:n_train])
-dtest = xgb.DMatrix(data[n_train:], label = label[n_train:])
-
-param = {'max_depth': 3, 'objective': 'binary:logistic', 'nthread': 10, 'eval_metric': 'auc'}
-
-num_round = 100
-
-evallist = [(dtest, 'eval'), (dtrain, 'train')]
-
-bst = xgb.train(param, dtrain, num_round, evallist)
-
-from sklearn.linear_model import LogisticRegression
-from sklearn.metrics import roc_auc_score
-
-model = LogisticRegression().fit(data[:n_train], label[:n_train])
-pred = model.predict_proba(data[n_train:])
-
-from ipdb import set_trace
-
-
-print roc_auc_score(label[n_train:].reshape(-1,), pred[:, 1].reshape(-1, ))
diff --git a/result/random_forest.py b/result/random_forest.py
new file mode 100644
index 00000000..a8778063
--- /dev/null
+++ b/result/random_forest.py
@@ -0,0 +1,27 @@
+import xgboost as xgb
+import numpy as np
+
+model_name = 'brits_i'
+
+impute = np.load('./{}_data.npy'.format(model_name)).reshape(-1, 48 * 35)
+label = np.load('./{}_label.npy'.format(model_name))
+
+data = np.nan_to_num(impute)
+
+n_train = 3000
+
+print(impute.shape)
+print(label.shape)
+
+from sklearn.ensemble import RandomForestClassifier
+from sklearn.metrics import roc_auc_score
+
+auc = []
+
+for i in range(10):
+    model = RandomForestClassifier().fit(data[:n_train], label[:n_train])
+    pred = model.predict_proba(data[n_train:])
+
+    auc.append(roc_auc_score(label[n_train:].reshape(-1,), pred[:, 1].reshape(-1, )))
+
+print(np.mean(auc))