removet tractate service feature

src/models/esmm/fe/tractate_fe.py

@@ -116,15 +116,15 @@ TRACTATE_COLUMNS = [
     "first_positions_num",
     "second_positions_num",
     "projects_num",
-    "is_related_service",
-    "effect_second_skip_num",
-    "business_second_skip_num",
-    "effect_second_skip_rate",
-    "business_second_skip_rate",
-    "service_id",
-    "service_price",
-    "service_sold_num",
-    "service_city",
+    # "is_related_service",
+    # "effect_second_skip_num",
+    # "business_second_skip_num",
+    # "effect_second_skip_rate",
+    # "business_second_skip_rate",
+    # "service_id",
+    # "service_price",
+    # "service_sold_num",
+    # "service_city",
     # "recommend_service_id",
     # "recommend_service_city",
     # "recommend_service_price",
@@ -197,10 +197,10 @@ INT_COLUMNS = [
     "first_positions_num",
     "second_positions_num",
     "projects_num",
-    "effect_second_skip_num",
-    "business_second_skip_num",
-    "service_price",
-    "service_sold_num",
+    # "effect_second_skip_num",
+    # "business_second_skip_num",
+    # "service_price",
+    # "service_sold_num",
 ]
 FLOAT_COLUMNS = [
     "one_ctr",
@@ -235,8 +235,8 @@ FLOAT_COLUMNS = [
     "sixty_browse_duration_avg",
     "ninety_browse_duration_avg",
     "history_browse_duration_avg",
-    "effect_second_skip_rate",
-    "business_second_skip_rate",
+    # "effect_second_skip_rate",
+    # "business_second_skip_rate",
 ]
 CATEGORICAL_COLUMNS = [
     "device_id",
@@ -290,9 +290,9 @@ CATEGORICAL_COLUMNS = [
     "click_tractate_id3",
     "click_tractate_id4",
     "click_tractate_id5",
-    "is_related_service",
-    "service_id",
-    "service_city",
+    # "is_related_service",
+    # "service_id",
+    # "service_city",
     # "recommend_service_id",
     # "recommend_service_city",
     # "recommend_service_price",
@@ -408,13 +408,13 @@ def tractate_feature_engineering(tractate_df):
     df["is_have_reply"] = df["is_have_reply"].astype(int)
     df["show_tag_id"] = df["show_tag_id"].astype(str)
 
-    df["is_related_service"] = df["is_related_service"].astype(int)
-    df["service_id"] = df["service_id"].astype(str)
+    # df["is_related_service"] = df["is_related_service"].astype(int)
+    # df["service_id"] = df["service_id"].astype(str)
     # df["recommend_service_id"] = df["recommend_service_id"].astype(str)
     # df["recommend_service_price"] = df["recommend_service_price"].astype(str)
 
-    df["service_id"] = df["service_city"].fillna("-1")
-    df["service_city"] = df["service_city"].fillna("")
+    # df["service_id"] = df["service_id"].fillna("-1")
+    # df["service_city"] = df["service_city"].fillna("")
     # df["recommend_service_id"] = df["recommend_service_id"].fillna("-1")
     # df["recommend_service_city"] = df["recommend_service_city"].fillna("")
 

src/models/esmm/model.py

@@ -28,18 +28,8 @@ def esmm_model_fn(features, labels, mode, params):
     cvr_logits = tf.layers.dense(last_cvr_layer, units=head.logits_dimension, kernel_initializer=tf.glorot_uniform_initializer())
     ctr_preds = tf.sigmoid(ctr_logits)
     cvr_preds = tf.sigmoid(cvr_logits)
-    ctcvr_preds = tf.multiply(ctr_preds, cvr_preds)
-    print("@@@@@@@@@@@")
-    print(ctr_logits)
-    print(ctr_preds)
-    print(ctcvr_preds)
-    print("@@@@@@@@@@@")
-    a = tf.multiply(0.8, ctr_preds)
-    b = tf.multiply(0.2, cvr_preds)
-    c = tf.multiply(a, b)
-    print(a)
-    print(c)
-    print("@@@@@@@@@@@")
+    # ctcvr_preds = tf.multiply(ctr_preds, cvr_preds)
+    ctcvr_preds = tf.multiply(tf.multiply(0.8, ctr_preds), tf.multiply(0.2, cvr_preds))
 
     # optimizer = tf.compat.v1.train.AdamOptimizer()
     # click_label = features["click_label"]

src/models/esmm/tractate_model.py

@@ -78,11 +78,11 @@ _int_columns = [
     "first_positions_num",
     "second_positions_num",
     "projects_num",
-    "is_related_service",
-    "effect_second_skip_num",
-    "business_second_skip_num",
-    "service_price",
-    "service_sold_num",
+    # "is_related_service",
+    # "effect_second_skip_num",
+    # "business_second_skip_num",
+    # "service_price",
+    # "service_sold_num",
 ]
 _float_columns = [
     "one_ctr",
@@ -117,8 +117,8 @@ _float_columns = [
     "sixty_browse_duration_avg",
     "ninety_browse_duration_avg",
     "history_browse_duration_avg",
-    "effect_second_skip_rate",
-    "business_second_skip_rate",
+    # "effect_second_skip_rate",
+    # "business_second_skip_rate",
 ]
 _categorical_columns = [
     "device_id",
@@ -166,8 +166,8 @@ _categorical_columns = [
     "click_tractate_id3",
     "click_tractate_id4",
     "click_tractate_id5",
-    "service_id",
-    "service_city",
+    # "service_id",
+    # "service_city",
     # "recommend_service_id",
     # "recommend_service_city",
     # "recommend_service_price",

src/train_tractate.py

@@ -62,7 +62,8 @@ def main():
     estimator_config = tf.estimator.RunConfig(session_config=session_config)
 
     model = tf.estimator.Estimator(model_fn=esmm_model_fn, params=params, model_dir=model_path, config=estimator_config)
-    train_spec = tf.estimator.TrainSpec(input_fn=lambda: esmm_input_fn(train_df, shuffle=True), max_steps=45000)
+    # TODO 45000
+    train_spec = tf.estimator.TrainSpec(input_fn=lambda: esmm_input_fn(train_df, shuffle=True), max_steps=12000)
     eval_spec = tf.estimator.EvalSpec(input_fn=lambda: esmm_input_fn(val_df, shuffle=False))
     res = tf.estimator.train_and_evaluate(model, train_spec, eval_spec)
     print("@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@")