-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain.py
161 lines (147 loc) · 7.85 KB
/
main.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
import data_preprocessing as dp
import pickle
import numpy as np
import pandas as pd
from sklearn.ensemble import RandomForestRegressor, GradientBoostingRegressor, VotingRegressor
from data_encoding import transform_categorial_into_numeric, transform_categorial_into_target
from hyperparamter_search import search_hyperparameter_random_forest, search_hyperparamter_gradientboosting
from machine_learning import fit_and_val_gradient_boosting, fit_and_val_random_forest, fit_and_val_ensemble_model
from settings import RF_PARA_GRID, GDB_PARA_GRID
def clean_house_prices_data():
df_train = dp.load_data("train.csv", encoding="utf-8")
df_train.drop(
labels=[17, 90, 102, 156, 182, 259, 342, 362, 371, 392, 520, 532, 533, 552, 646, 705, 736, 749, 778, 868,
894,
897, 984, 1000, 1011, 1035, 1045, 1048, 1049, 1090, 1179, 1216, 1218, 1230, 1321, 1412, 553,
1232, 39, 948, 332, 1379], axis=0)
# target_data = df_train.iloc[:, 80:81]
# del df_train["SalePrice"]
df_test = dp.load_data("test.csv", encoding="utf-8")
print(df_test.info())
df_test.insert(2, "SalePrice", [1] * df_test.shape[0], True)
df = pd.concat([df_train, df_test], keys=[0, 1])
# CLEAN DATA
# drop out columns that have more than 50% null objects over all data points or correlate with other features
dp.delete_features(df,
["Alley", "FireplaceQu", "PoolQC", "Fence", "MiscFeature", "Utilities", "Street",
"LowQualFinSF",
"GarageYrBlt", "BsmtFinType2"])
average = dp.replace_null_with_average_number(df, ["LotFrontage"])
df["MasVnrType"].replace(to_replace=np.nan, value="None", inplace=True)
df["MasVnrArea"].replace(to_replace=np.nan, value=0.0, inplace=True)
df["BsmtFullBath"].replace(to_replace=np.nan, value=0.0, inplace=True)
df["BsmtHalfBath"].replace(to_replace=np.nan, value=0.0, inplace=True)
df["BsmtFinSF1"].replace(to_replace=np.nan, value=0.0, inplace=True)
df["BsmtFinSF2"].replace(to_replace=np.nan, value=0.0, inplace=True)
df["BsmtUnfSF"].replace(to_replace=np.nan, value=0.0, inplace=True)
df["TotalBsmtSF"].replace(to_replace=np.nan, value=0.0, inplace=True)
df[['GarageQual', 'GarageCond', "GarageType", "GarageFinish", "BsmtQual", "BsmtCond", "BsmtExposure", "BsmtFinType1"]] = df[
["GarageQual", "GarageCond", "GarageType", "GarageFinish", "BsmtQual", "BsmtCond", "BsmtExposure", "BsmtFinType1"]].fillna('NI')
# drop rows which have missing values in choosen columns
df_train, df_test = df.xs(0), df.xs(1)
df_test.to_csv("modified_test.csv", encoding="utf-8")
# feature, data = dp.features_with_null_objects(df_test)
# data.to_csv("missing_data.csv", encoding="utf-8")
# data = dp.load_data("clean_test.csv", encoding="utf-8")
# print(data.info())
manual_changed_data = dp.load_data("manual_data.csv", encoding="utf-8", seperator=";")
dp.delete_features(manual_changed_data, ["Unnamed: 0"])
print(manual_changed_data.info())
manual_changed_data.set_index('Id')
df_test.set_index('Id')
print("test:")
print(df_test.info())
df_test.loc[manual_changed_data.index, :] = manual_changed_data[:] # ändert die Zeilen die manuel angepasst wurden aus missing_data.csv
print("Wichtig Wichtig")
print(df_test.info())
df_test = dp.load_data("final.csv", encoding="utf-8", seperator=";")
dp.delete_features(df_test, ["Unnamed: 0"])
df = pd.concat([df_train, df_test], keys=[0, 1])
# print(feature)
# print(data.info())
df = df.dropna(axis=0, subset=["MSZoning", "Exterior1st", "Exterior2nd", "BsmtFinSF1", "BsmtFinSF2", "BsmtUnfSF",
"TotalBsmtSF", "BsmtFullBath", "BsmtHalfBath", "KitchenQual", "Functional", "GarageCars",
"GarageArea", "SaleType", "BsmtQual", "BsmtExposure"])
return df
def encoding_data(df):
# transform categorial variables to a target(Encoding) for all possibilieties
df_train, df_test = df.xs(0), df.xs(1)
# df_train["SalePrice"].to_csv("check4", encoding="utf-8")
dp.delete_features(df_test, ["SalePrice"])
# df.insert(-1, "SalePrice", target_data, True)
features_transformed = ["MSZoning",
"LotShape", "LandContour", "LotConfig",
"LandSlope", "Neighborhood", "Condition1",
"Condition2",
"BldgType", "HouseStyle", "RoofStyle",
"RoofMatl",
"Exterior1st", "Exterior2nd", "MasVnrType",
"ExterQual",
"ExterCond", "Foundation", "BsmtQual",
"BsmtCond",
"BsmtExposure", "BsmtFinType1",
"Heating", "HeatingQC", "CentralAir",
"Electrical",
"KitchenQual", "Functional", "GarageType",
"GarageFinish", "GarageQual", "GarageCond",
"PavedDrive", "SaleType", "SaleCondition"]
df_train, categorial_encoders = transform_categorial_into_target(df_train, features_transformed)
for feature in features_transformed:
df_test[feature] = categorial_encoders[feature].transform(df_test[feature])
df_train.to_csv("clean_train_data.csv", encoding="utf-8")
df_test.to_csv("clean_test_data.csv", encoding="utf-8")
return df_train, df_test
def save_models(x_train, y_train, para_rf, para_gdb):
# save final model fit
rf = RandomForestRegressor(**para_rf)
rf.fit(x_train, y_train.values.ravel())
pickle.dump(rf, open("rf_model.sav", "wb"))
gdb = GradientBoostingRegressor(**para_gdb)
gdb.fit(x_train, y_train.values.ravel())
pickle.dump(gdb, open("gdb_model.pkl", "wb"))
rf = RandomForestRegressor(**para_rf)
gdb = GradientBoostingRegressor(**para_gdb)
ensemble_model = VotingRegressor([("rf", rf), ("grdb", gdb)], n_jobs=-1)
ensemble_model.fit(x_train, y_train.values.ravel())
pickle.dump(ensemble_model, open("models/ensemble_model.pkl", "wb"))
def final_predict(df_test):
print(df_test)
ensemble_model = pickle.load(open("models/ensemble_model.pkl", 'rb'))
id = df_test["Id"]
dp.delete_features(df_test, ["Id"])
print(df_test.info())
prediction = ensemble_model.predict(df_test)
# id = list(range(1461, 2920))
print(len(id))
print(len(prediction))
d = {'Id': id, 'SalePrice': prediction}
df = pd.DataFrame(data=d)
df.to_csv("submit.csv", index=False)
if __name__ == '__main__':
df = clean_house_prices_data()
df_train, df_test = encoding_data(df)
print("TestTest")
print(df_test.info())
dp.delete_features(df_train,
["Id"])
print("InfoInfo")
print(df_train.info())
x_train = df_train.iloc[:, 0:69]
print(x_train.info())
y_train = df_train.iloc[:, 69:70]
# para = search_hyperparamter_gradientboosting(x_train, y_train, "grid", GDB_PARA_GRID, plot_training=False)
# rf_model = search_hyperparameter_random_forest(x_train, y_train, "grid", RF_PARA_GRID, plot_training=True)
# print(para)
print(y_train)
# Parameter der Modelle die am vielversprechensten waren
para_rf = {"n_estimators": 50, "min_samples_split": 3, "min_samples_leaf": 1, "max_features": "sqrt",
"max_depth": None, "bootstrap": False}
para_gdb = {'subsample': 0.95, 'n_estimators': 500, 'min_samples_split': 3, 'min_samples_leaf': 2,
'max_features': 'auto',
'max_depth': 2, 'loss': 'squared_error', 'learning_rate': 0.05}
# fit_and_val_random_forest(x_train, y_train, para_rf)
# fit_and_val_gradient_boosting(x_train, y_train, para_gdb)
# fit_and_val_ensemble_model(x_train, y_train, para_rf, para_gdb)
# final model fit
# save_models(x_train, y_train, para_rf, para_gdb)
final_predict(df_test)