1. %load_ext autoreload
2. %autoreload 2
3. %matplotlib inline
4.  
5. from fastai.imports import *
6. from fastai.structured import *
7. import csv
8.  
9. from pandas_summary import DataFrameSummary
10. from sklearn.ensemble import RandomForestRegressor, RandomForestClassifier
11. from IPython.display import display
12.  
13. from sklearn import metrics
14.  
15. df_raw = pd.read_csv('Bulldozer_train.csv', low_memory=False, parse_dates=["saledate"])        
16.  
17. df_raw.SalePrice = np.log(df_raw.SalePrice)
18.  
19. add_datepart(df_raw, 'YearBuilt')
20. add_datepart(df_raw, 'YearRemodAdd')
21.  
22. train_cats(df_raw)
23.  
24. df, y, nas = proc_df(df_raw, 'SalePrice')
25.  
26. def split_vals(a,n): return a[:n].copy(), a[n:].copy()
27.  
28. n_valid = 140  # same as Kaggle's test set size
29. n_trn = len(df)-n_valid
30. raw_train, raw_valid = split_vals(df_raw, n_trn)
31. X_train, X_valid = split_vals(df, n_trn)
32. y_train, y_valid = split_vals(y, n_trn)
33.  
34. X_train.shape, y_train.shape, X_valid.shape
35.  
36. df_trn, y_trn, nas = proc_df(df_raw, 'SalePrice', subset=200, na_dict=nas)
37. X_train, _ = split_vals(df_trn, 150)
38. y_train, _ = split_vals(y_trn, 150)
39.  
40. m = RandomForestRegressor(n_estimators=40, min_samples_leaf=3, max_features=0.5, n_jobs=-1, oob_score=True)
41. m.fit(X_train, y_train)
42.  
43. df_test = pd.read_csv('Housing_test.csv', low_memory=False, parse_dates=["YearBuilt", "YearRemodAdd"])
44.  
45. add_datepart(df_test, 'YearBuilt')
46. add_datepart(df_test, 'YearRemodAdd')
47.  
48. train_cats(df_test)
49.  
50. df_1 = proc_df(df_test)
51.  
52. y_test= m.predict(df_1)
53.