import pandas as pd
import numpy as np

from sklearn.preprocessing import LabelEncoder

import gc

train = pd.read_csv('./train.csv')
test = pd.read_csv('./test.csv')

data_info=pd.read_csv('data_info.csv')
data_info=pd.concat([data_info[:4],data_info[5:]])
data_info=data_info.reset_index()[['변수명','변수 설명']]
data_info

train

train.describe()

import seaborn as sns

# id 제외
df=train.drop(columns='id')
#요일 전처리
df['day_of_week']=df['day_of_week'].map({'월':1,'화':2,'수':3,'목':4,'금':5,'토':6,'일':7})
df

train[train.describe().columns]

ind=train.describe().columns.drop(['base_date','start_latitude','start_longitude','end_latitude','end_longitude'])
ind

    Index(['base_hour', 'lane_count', 'road_rating', 'multi_linked',
           'connect_code', 'maximum_speed_limit', 'vehicle_restricted',
           'weight_restricted', 'height_restricted', 'road_type', 'target'],
          dtype='object')

train[ind].describe()

train.loc[:500,ind]

import matplotlib.pyplot as plt
plt.figure(figsize=(10,10))
sns.heatmap(train.loc[:500,ind].astype(float).corr(), linewidths = 0.1, vmax = 1.0, fmt=".1f",
           square = True, cmap = 'Blues', linecolor = "white", annot = True, annot_kws = {"size" : 16})

str_col = ['day_of_week','start_turn_restricted','end_turn_restricted']
for i in str_col:
    le = LabelEncoder()
    le=le.fit(train[i])
    train[i]=le.transform(train[i])
    
    for label in np.unique(test[i]):
        if label not in le.classes_: 
            le.classes_ = np.append(le.classes_, label)
    test[i]=le.transform(test[i])

train=train.loc[:2700000]
valid=train.loc[2700000:]

y_train = train['target'] 

X_train = train.drop(['id','base_date', 'target','road_name', 'start_node_name', 'end_node_name','vehicle_restricted'], axis=1)

y_valid = valid['target'] 

X_valid = valid.drop(['id','base_date', 'target','road_name', 'start_node_name', 'end_node_name','vehicle_restricted'], axis=1)


test = test.drop(['id','base_date', 'road_name', 'start_node_name', 'end_node_name','vehicle_restricted'], axis=1)

print(X_train.shape)
print(y_train.shape)
print(test.shape)

    (2700001, 16)
    (2700001,)
    (291241, 16)

LR = lgb.LGBMRegressor(random_state=42).fit(X_train, y_train)
pred = LR.predict(X_valid)

from sklearn.metrics import mean_absolute_error
mae = mean_absolute_error(y_valid,pred)
print(mae)

    1.7515819641866202

from sklearn.tree import DecisionTreeRegressor

tree = DecisionTreeRegressor(random_state=123).fit(X_train, y_train)
pred = tree.predict(X_valid)

from sklearn.metrics import mean_absolute_error
mae = mean_absolute_error(y_valid,pred)
print(mae)

    2.6666666666666643

from sklearn.ensemble import RandomForestRegressor
rf=RandomForestRegressor().fit(X_train,y_train)
pred = rf.predict(X_valid)

from sklearn.metrics import mean_absolute_error
mae = mean_absolute_error(y_valid,pred)
print(mae)

    2.6569172934459857

grid = {
    'n_estimators' : [100,200],
    'max_depth' : [6,8,10,12],
    'min_samples_leaf' : [3,5,7,10],
    'min_samples_split' : [2,3,5,10]
}

from sklearn.model_selection import GridSearchCV
classifier_grid = GridSearchCV(LR, param_grid = grid, scoring="accuracy", n_jobs=-1, verbose =1)

classifier_grid.fit(X_train, y_train)

print("최고 평균 정확도 : {}".format(classifier_grid.best_score_))
print("최고의 파라미터 :", classifier_grid.best_params_)

    Fitting 5 folds for each of 128 candidates, totalling 640 fits


    /opt/homebrew/Caskroom/miniforge/base/lib/python3.9/site-packages/sklearn/model_selection/_search.py:969: UserWarning: One or more of the test scores are non-finite: [nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
     nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
     nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
     nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
     nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
     nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
     nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan nan
     nan nan]
      warnings.warn(


    [LightGBM] [Warning] Unknown parameter: min_samples_leaf
    [LightGBM] [Warning] Unknown parameter: min_samples_split
    [LightGBM] [Warning] Accuracy may be bad since you didn't explicitly set num_leaves OR 2^max_depth > num_leaves. (num_leaves=31).
    최고 평균 정확도 : nan
    최고의 파라미터 : {'max_depth': 6, 'min_samples_leaf': 3, 'min_samples_split': 2, 'n_estimators': 100}

grid = {
    'n_estimators' : [100,200],
    'max_depth' : [6,8,10],
    'min_samples_leaf' : [3,5,7],
    'min_samples_split' : [2,3,5]
}

from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import RandomizedSearchCV
rf=RandomForestRegressor().fit(X_train,y_train)
rand = RandomizedSearchCV(rf, grid, n_jobs=-1, cv=5, n_iter=3, random_state=1) 
rand_result = rand.fit(X_train, y_train)
print(rand_result.best_score_)
print(rand_result.best_params_)

    0.809383983301521
    {'n_estimators': 100, 'min_samples_split': 3, 'min_samples_leaf': 5, 'max_depth': 10}

LR = lgb.LGBMRegressor(random_state=42).fit(X_train, y_train)
from sklearn.model_selection import RandomizedSearchCV
rand = RandomizedSearchCV(LR, grid, n_jobs=-1, cv=5, n_iter=3, random_state=1) 
rand_result = rand.fit(X_train, y_train)
print(rand_result.best_score_)
print(rand_result.best_params_)

    0.8682401067973238
    {'n_estimators': 200, 'min_samples_split': 2, 'min_samples_leaf': 10, 'max_depth': 12}

#데이터 로드
train = pd.read_csv('./train.csv')
test = pd.read_csv('./test.csv')
# 데이터 전처리
str_col = ['day_of_week','start_turn_restricted','end_turn_restricted']
for i in str_col:
    le = LabelEncoder()
    le=le.fit(train[i])
    train[i]=le.transform(train[i])
    
    for label in np.unique(test[i]):
        if label not in le.classes_: 
            le.classes_ = np.append(le.classes_, label)
    test[i]=le.transform(test[i])
#훈련, 테스트 데이터 분리
y_train = train['target'] 
X_train = train.drop(['id','base_date', 'target','road_name', 'start_node_name', 'end_node_name','vehicle_restricted'], axis=1)
test = test.drop(['id','base_date', 'road_name', 'start_node_name', 'end_node_name','vehicle_restricted'], axis=1)

#모델 훈련 & 예측
LR = lgb.LGBMRegressor(n_estimators=100,min_samples_split=2, min_samples_leaf=3, max_depth=6).fit(X_train,y_train)
pred = LR.predict(test)

sample_submission = pd.read_csv('sample_submission.csv')

sample_submission['target'] = pred
sample_submission.to_csv("./submit.csv", index = False)

sample_submission