近期文章导入数据
import pandas as pd
df = pd.read_csv('data/train.csv')
df.head()
我们看一下季节和假期对出行对影响,用到seaborn库对可视化。 import matplotlib.pyplot as plt
import seaborn as sns
%matplotlib inline
grid = sns.FacetGrid(df, row='holiday', col='season')
grid.map(plt.scatter, 'temp', 'casual', alpha=0.4)
grid.add_legend()
从上面可以看出季节对出行影响较少,但是否为节假日对出行影响很大。 数据预处理把datetime列数据变为pd.datetime类型后,可以使用其特殊对日期数据操作。 我们将日期分为月、日、时三个列。 我们对温度、湿度数据进行标准正态化处理,最后剔除掉datetime列。 df['datetime'] = df['datetime'].apply(pd.to_datetime)
#df['month'] = df['datetime'].dt.month
df['month'] = df['datetime'].apply(lambda x:x.month)
df['day'] = df['datetime'].apply(lambda x:x.day)
df['hour'] = df['datetime'].apply(lambda x:x.hour)
df['temp'] = (df['temp']-df['temp'].mean())/df['temp'].std()
df['atemp'] = (df['atemp']-df['atemp'].mean())/df['atemp'].std()
df['humidity'] = (df['humidity']-df['humidity'].mean())/df['humidity'].std()
df.drop(['datetime'], axis=1, inplace=True)
df.sample(5)
X = df[['season', 'holiday', 'workingday', 'weather', 'temp', 'atemp', 'humidity', 'windspeed', 'month', 'day', 'hour']]
y = df['count']
数据分割将数据按照三七开,三分测试数据,七分训练数据。 from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=2019)
训练模型这里因为我们对y是连续型变量,所以我们做的不是分类模型,而是回归模型。这里选了DecisionTreeRegressor、RandomForestRegressor分别测试 from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import RandomForestRegressor
dtr = DecisionTreeRegressor()
dtr.fit(X_train, y_train)
dtr.score(X_test, y_test)
0.7334594631440063
DecisionTreeRegressor决策树模型准确率达到73%,还不错。再看看RandomForestRegressor表现如何 rfr = RandomForestRegressor(n_estimators=200)
rfr.fit(X_train, y_train)
rfr.score(X_test, y_test)
0.8639249009469554
没怎么清洗数据、调参数炼丹药,RandomForestRegressor模型准确率就能达到86%,棒棒哒~
|