天池大赛之工业蒸汽处理(改进版 ---- 0.1235）

导包

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
%matplotlib inline
import seaborn as sns
from sklearn.linear_model import LinearRegression,Lasso,Ridge,ElasticNet,RidgeCV
from sklearn.neighbors import KNeighborsRegressor
from sklearn.ensemble import GradientBoostingRegressor,RandomForestRegressor,AdaBoostRegressor,ExtraTreesRegressor
from xgboost import XGBRegressor
from lightgbm import LGBMRegressor
#支持向量机
from sklearn.svm import SVR
from sklearn.metrics import mean_squared_error
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import MinMaxScaler,StandardScaler,PolynomialFeatures
import warnings
warnings.filterwarnings('ignore')

数据聚合

train_data = pd.read_csv('./zhengqi_train.txt',sep='\t')
test_data = pd.read_csv('./zhengqi_test.txt',sep='\t')

#合并训练数据和预测数据
train_data["origin"]="train"
test_data["origin"]="test"
data_all=pd.concat([train_data,test_data],axis=0,ignore_index=True)
#View data
data_all

	V0	V1	V10	V11	V12	V13	V14	V15	V16	V17	V18	V19	V2	V20	V21	V22	V23	V24	V25	V26	V27	V28	V29	V3	V30	V31	V32	V33	V34	V35	V36	V37	V4	V5	V6	V7	V8	V9	origin	target
0	0.566	0.016	-0.940	-0.307	-0.073	0.550	-0.484	0.000	-1.707	-1.162	-0.573	-0.991	-0.143	0.610	-0.400	-0.063	0.356	0.800	-0.223	0.796	0.168	-0.450	0.136	0.407	0.109	-0.615	0.327	-4.627	-4.789	-5.101	-2.608	-3.508	0.452	-0.901	-1.812	-2.360	-0.436	-2.114	train	0.175
1	0.968	0.437	0.188	-0.455	-0.134	1.109	-0.488	0.000	-0.977	-1.162	-0.571	-0.836	0.066	0.588	-0.802	-0.063	0.357	0.801	-0.144	1.057	0.338	0.671	-0.128	0.566	0.124	0.032	0.600	-0.843	0.160	0.364	-0.335	-0.730	0.194	-0.893	-1.566	-2.360	0.332	-2.114	train	0.676
2	1.013	0.568	0.874	-0.051	-0.072	0.767	-0.493	-0.212	-0.618	-0.897	-0.564	-0.558	0.235	0.576	-0.477	-0.063	0.355	0.961	-0.067	0.915	0.326	1.287	-0.009	0.370	0.361	0.277	-0.116	-0.843	0.160	0.364	0.765	-0.589	0.112	-0.797	-1.367	-2.360	0.396	-2.114	train	0.633
3	0.733	0.368	0.011	0.102	-0.014	0.769	-0.371	-0.162	-0.429	-0.897	-0.574	-0.564	0.283	0.272	-0.491	-0.063	0.352	1.435	0.113	0.898	0.277	1.298	0.015	0.165	0.417	0.279	0.603	-0.843	-0.065	0.364	0.333	-0.112	0.599	-0.679	-1.200	-2.086	0.403	-2.114	train	0.206
4	0.684	0.638	-0.251	0.570	0.199	-0.349	-0.342	-0.138	-0.391	-0.897	-0.572	-0.394	0.260	0.106	0.309	-0.259	0.352	0.881	0.221	0.386	0.332	1.289	0.183	0.209	1.078	0.328	0.418	-0.843	-0.215	0.364	-0.280	-0.028	0.337	-0.454	-1.073	-2.086	0.314	-2.114	train	0.384
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
4808	-1.362	-1.553	-2.551	0.518	0.396	0.928	1.452	0.867	-5.143	1.227	-3.573	0.107	-3.096	-0.088	0.227	2.953	-1.538	-0.630	-3.072	-1.120	-1.674	0.525	0.171	-0.444	-4.488	-5.793	-4.050	-1.187	-0.852	-2.131	-2.564	0.597	0.381	1.375	-4.854	-5.331	-4.074	-3.838	test	NaN
4809	-2.698	-3.452	-2.525	0.311	-1.786	1.871	1.885	1.135	-5.774	1.227	-0.965	0.193	-3.620	-0.506	-0.574	3.149	-1.479	-0.204	-3.432	-2.101	-1.773	-0.446	1.297	-1.066	-0.613	-7.698	-0.674	-1.187	-0.852	-2.131	-2.564	1.215	-1.385	1.378	-4.927	-5.103	-4.393	-1.683	test	NaN
4810	-2.615	-3.564	-2.529	-0.029	-1.151	1.976	2.337	0.504	-4.752	1.492	-1.568	0.301	-3.402	0.109	-0.541	3.511	-1.085	1.057	-2.409	0.477	-1.585	-0.447	0.552	-0.422	0.125	-6.111	0.275	-1.851	-1.548	-1.537	-2.544	1.612	-1.272	1.121	-4.223	-4.315	-5.196	-3.407	test	NaN
4811	-2.661	-3.646	-2.560	-0.028	-1.512	1.520	2.243	0.206	-4.200	1.492	-1.282	-0.036	-3.271	-1.015	-0.203	3.511	-1.084	0.800	-2.339	0.050	-1.410	-0.447	0.318	-0.699	1.086	-5.268	0.683	-1.645	-1.471	-1.537	-2.549	1.431	-1.270	1.116	-3.716	-3.809	-4.735	-2.976	test	NaN
4812	-2.321	-3.037	0.056	0.306	-1.154	0.847	2.221	0.206	-3.960	1.492	-1.213	0.592	-3.214	-1.502	0.153	3.609	-1.088	0.799	-2.339	-0.077	-1.242	-0.442	0.323	-1.594	-0.774	-5.211	1.618	-1.703	-1.471	-1.537	-1.123	1.988	-0.910	1.259	-3.616	-3.747	-4.368	-2.976	test	NaN

4813 rows × 40 columns

特征探索

#探索出去最后两列的数字属性
data_all.columns[:-2]

Index(['V0', 'V1', 'V10', 'V11', 'V12', 'V13', 'V14', 'V15', 'V16', 'V17',
       'V18', 'V19', 'V2', 'V20', 'V21', 'V22', 'V23', 'V24', 'V25', 'V26',
       'V27', 'V28', 'V29', 'V3', 'V30', 'V31', 'V32', 'V33', 'V34', 'V35',
       'V36', 'V37', 'V4', 'V5', 'V6', 'V7', 'V8', 'V9'],
      dtype='object')

#38个特征将一些不重要的删除
#特征分布情况，训练和测试数据特征分布不均匀，删除
for column in data_all.columns[0:-2]:
    g = sns.kdeplot(data_all[column][(data_all["origin"] == "train")], color="Red", shade = True)
    g = sns.kdeplot(data_all[column][(data_all["origin"] == "test")], ax =g, color="Blue", shade= True)
    g.set_xlabel(column)
    g.set_ylabel("Frequency")
    g = g.legend(["train","test"])
    plt.show()

fig = plt.figure(figsize=(10, 10))
for i in range(len(data_all.columns)-2):
    g = sns.FacetGrid(data_all, col='origin')
    g = g.map(sns.distplot, data_all.columns[i])

<Figure size 720x720 with 0 Axes>

#通过图示可以看出'V11','V17','V22','V5'，波动太大，删除
drop_labels = ['V11','V17','V22','V5']
data_all.drop(drop_labels,axis=1,inplace=True)

相关性系数corr

# 找出相关程度
plt.figure(figsize=(20, 16))  # 指定绘图对象宽度和高度
mcorr = train_data.corr()  # 相关系数矩阵，即给出了任意两个变量之间的相关系数
mask = np.zeros_like(mcorr, dtype=np.bool)  # 构造与mcorr同维数矩阵 为bool型
mask[np.triu_indices_from(mask)] = True  # 角分线右侧为True
cmap = sns.diverging_palette(220, 10, as_cmap=True)  # 返回matplotlib colormap对象
g = sns.heatmap(mcorr, mask=mask, cmap=cmap, square=True, annot=True, fmt='0.2f')  # 热力图（看两两相似度）
plt.show

<function matplotlib.pyplot.show(*args, **kw)>

在这里插入图片描述

# 通过相关性系数找到7个相关性不大的属性
cond = mcorr.loc['target'].abs()<0.1
drop_labels = mcorr.loc['target'][cond].index
#['V14', 'V21', 'V25', 'V26', 'V32', 'V33', 'V34']

#查看属性分布后，将分布不好的删除  （'V14', 'V21', ）
drop_labels = ['V14', 'V21']
data_all.drop(drop_labels,axis=1,inplace=True)

#删除了6个属性
data_all.shape

(4813, 34)

对数据进行归一化

data = data_all.iloc[:,:-2]
minmaxscale = MinMaxScaler()
data = minmaxscale.fit_transform(data)
data

array([[0.77577505, 0.723449  , 0.22174265, ..., 0.43285165, 0.66410771,
        0.73528007],
       [0.83374189, 0.77878549, 0.37388724, ..., 0.43285165, 0.7548128 ,
        0.73528007],
       [0.84023071, 0.79600421, 0.46641489, ..., 0.43285165, 0.76237156,
        0.73528007],
       ...,
       [0.31708724, 0.25289169, 0.0074184 , ..., 0.17367095, 0.10192512,
        0.64706284],
       [0.31045422, 0.24211356, 0.00323712, ..., 0.24075302, 0.1563718 ,
        0.67646858],
       [0.35948089, 0.32216088, 0.35608309, ..., 0.24897256, 0.19971655,
        0.67646858]])

#归一化数据
data_all_norm = pd.DataFrame(data,columns=data_all.columns[:-2])
data_all_norm

	V0	V1	V10	V12	V13	V15	V16	V18	V19	V2	V20	V23	V24	V25	V26	V27	V28	V29	V3	V30	V31	V32	V33	V34	V35	V36	V37	V4	V6	V7	V8	V9
0	0.775775	0.723449	0.221743	0.570828	0.694786	0.402245	0.487950	0.375125	0.380238	0.582197	0.537946	0.792169	0.569153	0.375250	0.730736	0.902936	0.279341	0.406834	0.665193	0.603714	0.729379	0.679479	0.000000	0.000000	0.242424	0.000000	0.018343	0.571839	0.508616	0.432852	0.664108	0.735280
1	0.833742	0.778785	0.373887	0.564418	0.778544	0.402245	0.569779	0.375374	0.401962	0.611588	0.534996	0.792304	0.569419	0.381824	0.762915	0.924734	0.437095	0.371596	0.689434	0.605676	0.796005	0.721792	0.374950	0.499949	0.800020	0.289702	0.436025	0.544381	0.541225	0.432852	0.754813	0.735280
2	0.840231	0.796004	0.466415	0.570933	0.727300	0.372870	0.610021	0.376246	0.440925	0.635354	0.533387	0.792035	0.611893	0.388232	0.745407	0.923195	0.523783	0.387480	0.659552	0.636673	0.821234	0.610818	0.374950	0.499949	0.800020	0.429901	0.457224	0.535653	0.567603	0.432852	0.762372	0.735280
3	0.799856	0.769716	0.350013	0.577028	0.727600	0.379798	0.631207	0.375000	0.440084	0.642104	0.492625	0.791633	0.737722	0.403212	0.743312	0.916912	0.525331	0.390683	0.628297	0.643997	0.821440	0.722257	0.374950	0.477220	0.800020	0.374841	0.528943	0.587484	0.589740	0.469177	0.763198	0.735280
4	0.792790	0.805205	0.314675	0.599412	0.560084	0.383123	0.635467	0.375249	0.463910	0.638869	0.470367	0.791633	0.590656	0.412200	0.680187	0.923965	0.524064	0.413107	0.635005	0.730447	0.826485	0.693583	0.374950	0.462067	0.800020	0.296712	0.541573	0.559600	0.606575	0.469177	0.752687	0.735280
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
4808	0.497765	0.517219	0.004451	0.620113	0.751423	0.522378	0.102791	0.001246	0.534128	0.166924	0.444355	0.538214	0.189541	0.138149	0.494514	0.666752	0.416549	0.411506	0.535447	0.002485	0.196169	0.001085	0.340864	0.397717	0.545455	0.005608	0.635544	0.564283	0.105382	0.038977	0.234440	0.617657
4809	0.305119	0.267613	0.007958	0.390815	0.892718	0.559512	0.032059	0.326271	0.546181	0.093236	0.388308	0.546125	0.302628	0.108189	0.373567	0.654058	0.279904	0.561799	0.440616	0.509286	0.000000	0.524334	0.340864	0.397717	0.545455	0.005608	0.728462	0.376330	0.095705	0.069203	0.196764	0.764686
4810	0.317087	0.252892	0.007418	0.457545	0.908451	0.472080	0.146620	0.251122	0.561317	0.123893	0.470770	0.598954	0.637377	0.193326	0.691407	0.678164	0.279764	0.462360	0.538802	0.605807	0.163423	0.671420	0.275069	0.327407	0.606061	0.008157	0.788152	0.388357	0.189024	0.173671	0.101925	0.647063
4811	0.310454	0.242114	0.003237	0.419609	0.840126	0.430788	0.208497	0.286765	0.514085	0.142315	0.320059	0.599088	0.569153	0.199151	0.638762	0.700603	0.279764	0.431127	0.496570	0.731494	0.250232	0.734656	0.295482	0.335185	0.606061	0.007520	0.760938	0.388570	0.256230	0.240753	0.156372	0.676469
4812	0.359481	0.322161	0.356083	0.457230	0.739287	0.430788	0.235400	0.295364	0.602102	0.150330	0.254760	0.598552	0.568888	0.199151	0.623104	0.722144	0.280467	0.431794	0.360116	0.488229	0.256101	0.879572	0.289734	0.335185	0.606061	0.189268	0.844685	0.426884	0.269486	0.248973	0.199717	0.676469

4813 rows × 32 columns

#将oringin和target属性merage上
data_all_norm = pd.merge(data_all_norm,data_all.iloc[:,-2:],left_index=True,right_index=True)

data_all_norm.describe()

	V0	V1	V10	V12	V13	V15	V16	V18	V19	V2	V20	V23	V24	V25	V26	V27	V28	V29	V3	V30	V31	V32	V33	V34	V35	V36	V37	V4	V6	V7	V8	V9	target
count	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	4813.000000	2888.000000
mean	0.694172	0.721357	0.348518	0.578507	0.612372	0.402251	0.679294	0.446542	0.519158	0.602300	0.456147	0.744438	0.356712	0.393796	0.632582	0.881401	0.342653	0.388683	0.603139	0.589459	0.792709	0.628824	0.458493	0.483790	0.762873	0.332385	0.545795	0.523743	0.748823	0.745740	0.715607	0.879536	0.126353
std	0.144198	0.131443	0.134882	0.105088	0.149835	0.138561	0.112095	0.124627	0.140166	0.140628	0.134083	0.134085	0.265512	0.083226	0.123294	0.128221	0.140731	0.133475	0.152462	0.130786	0.102976	0.155003	0.099095	0.101020	0.102037	0.127456	0.150356	0.106430	0.132560	0.132577	0.118105	0.068244	0.983966
min	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	0.000000	-3.044000
25%	0.626676	0.679416	0.284327	0.532892	0.519928	0.299016	0.629414	0.399302	0.414436	0.514414	0.370475	0.719362	0.040616	0.347870	0.566515	0.888575	0.278778	0.292445	0.503888	0.550092	0.761816	0.562461	0.409037	0.454490	0.727273	0.270584	0.445647	0.478182	0.683324	0.696938	0.664934	0.852903	-0.350250
50%	0.729488	0.752497	0.366469	0.591635	0.627809	0.391437	0.700258	0.456256	0.540294	0.617072	0.447305	0.788817	0.381736	0.388815	0.641228	0.916015	0.279904	0.375734	0.614270	0.594428	0.815055	0.643056	0.454518	0.499949	0.800020	0.347056	0.539317	0.535866	0.774125	0.771974	0.742884	0.882377	0.313000
75%	0.790195	0.799553	0.432965	0.641971	0.719958	0.489954	0.753279	0.501745	0.623125	0.700464	0.522660	0.792706	0.574728	0.427597	0.713599	0.932555	0.413031	0.471837	0.710474	0.650798	0.852229	0.719777	0.500000	0.511365	0.800020	0.414861	0.643061	0.585036	0.842259	0.836405	0.790835	0.941189	0.793250
max	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	1.000000	2.538000

def scale_minmax(data):
    return (data - data.min())/(data.max() - data.min())

#使用Box-Cox将连续数据转换的更加平滑（主要处理类似正太分布）
from scipy import stats
fcols = 6
frows = len(data_all_norm.columns[:10])
plt.figure(figsize=(4*fcols,4*frows))
i = 0

for col in data_all_norm.columns[:10]:
    dat = data_all_norm[[col, 'target']].dropna()

#     这条线就是数据分布dist：distribution（分布）
    i+=1
    plt.subplot(frows,fcols,i)
    sns.distplot(dat[col],fit = stats.norm);
    plt.title(col+' Original')
    plt.xlabel('')

#     第二个图：skew统计分析中中一个属性
#     skewness 偏斜系数，对正太分布的度量
    i+=1
    plt.subplot(frows,fcols,i)
    _=stats.probplot(dat[col], plot=plt)#画图，偏析度
    plt.title('skew='+'{:.4f}'.format(stats.skew(dat[col])))
    plt.xlabel('')
    plt.ylabel('')

#     散点图
    i+=1
    plt.subplot(frows,fcols,i)
#     plt.plot(dat[var], dat['target'],'.',alpha=0.5)
    plt.scatter(dat[col],dat['target'],alpha=0.5)
    plt.title('corr='+'{:.2f}'.format(np.corrcoef(dat[col], dat['target'])[0][1]))

#     ！！！对数据进行了处理！！！
#   数据分布图distribution
    i+=1
    plt.subplot(frows,fcols,i)
    trans_var, lambda_var = stats.boxcox(dat[col].dropna()+1)
    trans_var = scale_minmax(trans_var)      
    sns.distplot(trans_var , fit=stats.norm);
    plt.title(col+' Tramsformed')
    plt.xlabel('')

#     偏斜度
    i+=1
    plt.subplot(frows,fcols,i)
    _=stats.probplot(trans_var, plot=plt)
    plt.title('skew='+'{:.4f}'.format(stats.skew(trans_var)))
    plt.xlabel('')
    plt.ylabel('')

#     散点图
    i+=1
    plt.subplot(frows,fcols,i)
    plt.plot(trans_var, dat['target'],'.',alpha=0.5)
    plt.title('corr='+'{:.2f}'.format(np.corrcoef(trans_var,dat['target'])[0][1]))

# 将数据进行Box-Cox转换
# 统计建模中常用的数据变化
# 数据更加正态化，标准化
for col in data_all_norm.columns[:-2]:
    boxcox,maxlog = stats.boxcox(data_all_norm[col] + 1)
    data_all_norm[col] = scale_minmax(boxcox)

data_all_norm

	V0	V1	V10	V12	V13	V15	V16	V18	V19	V2	V20	V23	V24	V25	V26	V27	V28	V29	V3	V30	V31	V32	V33	V34	V35	V36	V37	V4	V6	V7	V8	V9	origin	target
0	0.507483	0.357070	0.134959	0.303471	0.561751	0.539735	0.136013	0.239798	0.272914	0.442658	0.694629	0.425929	0.592470	0.626176	0.552721	0.394651	0.377657	0.559002	0.581357	0.323667	0.267157	0.440715	0.000000	0.000000	0.026476	0.000000	0.020896	0.353680	0.165759	0.094056	0.304061	0.253539	train	0.175
1	0.610419	0.445015	0.253597	0.297055	0.668704	0.539735	0.197424	0.240004	0.292263	0.474668	0.692110	0.426178	0.592730	0.632878	0.597390	0.488267	0.547131	0.522334	0.608782	0.325784	0.376810	0.496931	0.355631	0.432280	0.467466	0.175457	0.466089	0.325746	0.190998	0.094056	0.430326	0.253539	train	0.676
2	0.622895	0.475812	0.336900	0.303577	0.602125	0.509062	0.234823	0.240729	0.328031	0.501306	0.690732	0.425680	0.634207	0.639320	0.572815	0.481023	0.630446	0.539067	0.575040	0.360554	0.427728	0.359039	0.355631	0.432280	0.467466	0.290367	0.487365	0.317154	0.213520	0.094056	0.442528	0.253539	train	0.633
3	0.548433	0.429476	0.233498	0.309766	0.602504	0.516390	0.256694	0.239694	0.327245	0.508997	0.654865	0.424935	0.755301	0.654039	0.569916	0.452488	0.631879	0.542400	0.540475	0.369141	0.428167	0.497575	0.355631	0.409624	0.467466	0.242724	0.558547	0.370220	0.233980	0.113283	0.443879	0.253539	train	0.206
4	0.536158	0.492987	0.204775	0.333233	0.409379	0.519887	0.261284	0.239901	0.349778	0.505305	0.634468	0.424935	0.613508	0.662648	0.486902	0.484632	0.630707	0.565360	0.547831	0.482005	0.439062	0.458936	0.355631	0.394675	0.467466	0.180709	0.570959	0.341058	0.250548	0.113283	0.426944	0.253539	train	0.384
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
4808	0.184452	0.144638	0.002294	0.356001	0.633015	0.655194	0.009837	0.000579	0.419213	0.092622	0.609861	0.127612	0.205965	0.302314	0.286901	0.033158	0.526439	0.563741	0.442151	0.000452	0.007696	0.000245	0.322227	0.332586	0.152596	0.002675	0.662217	0.345855	0.012300	0.002622	0.035277	0.128271	test	NaN
4809	0.074357	0.036169	0.004113	0.155465	0.830440	0.687976	0.002518	0.200542	0.431591	0.048667	0.553814	0.132960	0.324364	0.246144	0.187745	0.028743	0.378309	0.702706	0.348424	0.232703	0.000000	0.271468	0.322227	0.332586	0.152596	0.002675	0.750654	0.182800	0.010893	0.005091	0.026747	0.298339	test	NaN
4810	0.079352	0.032715	0.003833	0.203011	0.854218	0.608691	0.015871	0.145112	0.447329	0.066348	0.634842	0.173746	0.658944	0.395001	0.501062	0.037668	0.378146	0.613591	0.445590	0.325925	0.005566	0.430530	0.258323	0.267339	0.203097	0.003899	0.806584	0.191546	0.027408	0.017464	0.010723	0.152833	test	NaN
4811	0.076558	0.030332	0.001667	0.175014	0.753601	0.568571	0.026862	0.170668	0.398930	0.077385	0.479359	0.173862	0.592470	0.404054	0.436769	0.048282	0.378146	0.583346	0.402911	0.483508	0.012412	0.514964	0.278067	0.274424	0.203097	0.003592	0.781167	0.191703	0.044201	0.029609	0.019058	0.181492	test	NaN
4812	0.098781	0.051178	0.238555	0.202767	0.617374	0.568571	0.032709	0.177030	0.490804	0.082286	0.400802	0.173400	0.592210	0.404054	0.418681	0.061087	0.378960	0.584005	0.274131	0.215143	0.013031	0.751303	0.272500	0.274424	0.203097	0.105645	0.858952	0.221029	0.048101	0.031384	0.027365	0.181492	test	NaN

4813 rows × 34 columns

过滤异常值

ridge = RidgeCV(alphas=[0.0001,0.001,0.01,0.1,0.2,0.5,1,2,3,4,5,10,20,30,50])

cond = data_all_norm['origin'] == 'train'

X_train = data_all_norm[cond].iloc[:,:-2]
# 真实值
y_train = data_all_norm[cond]['target']
# 算法拟合数据和目标值的时候，不可能100%拟合
ridge.fit(X_train,y_train)
# 预测，预测值肯定会和真实值有一定的偏差，偏差特别大，当成异常值
y_ = ridge.predict(X_train)

cond = abs(y_ - y_train) > y_train.std()
print(cond.sum())
# 画图
plt.figure(figsize=(12,6))
axes = plt.subplot(1,3,1)
axes.scatter(y_train,y_)
axes.scatter(y_train[cond],y_[cond],c = 'red',s = 20)

axes = plt.subplot(1,3,2)
axes.scatter(y_train,y_train - y_)
axes.scatter(y_train[cond],(y_train - y_)[cond],c = 'red')

axes = plt.subplot(1,3,3)
# _ = axes.hist(y_train,bins = 50)
(y_train - y_).plot.hist(bins = 50,ax = axes)
(y_train - y_).loc[cond].plot.hist(bins = 50,ax = axes,color = 'r')

40





<matplotlib.axes._subplots.AxesSubplot at 0x2403c0836a0>

index = cond[cond].index

data_all_norm.drop(index,axis = 0,inplace=True)

cond = data_all_norm['origin'] == 'train'
X_train = data_all_norm[cond].iloc[:,:-2]
y_train = data_all_norm[cond]['target']

cond = data_all_norm['origin'] == 'test'
X_test = data_all_norm[cond].iloc[:,:-2]

使用不同算法进行计算，最后求取平均值！！

estimators = {}
estimators['forest'] = RandomForestRegressor(n_estimators=300)
estimators['gbdt'] = GradientBoostingRegressor(n_estimators=300)
estimators['ada'] = AdaBoostRegressor(n_estimators=300)
estimators['extreme'] = ExtraTreesRegressor(n_estimators=300)
estimators['svm_rbf'] = SVR(kernel='rbf')
estimators['light'] = LGBMRegressor(n_estimators=300)
estimators['xgb'] = XGBRegressor(n_estimators=300)

#将结果存入列表中，求取平均值作为最后答案
result = []
for key,model in estimators.items():
    model.fit(X_train,y_train)
    y_ = model.predict(X_test)
    result.append(y_)
y_ = np.mean(result,axis = 0)

pd.Series(y_).to_csv('./norm.txt',index = False)

[19:51:26] WARNING: C:/Jenkins/workspace/xgboost-win64_release_0.90/src/objective/regression_obj.cu:152: reg:linear is now deprecated in favor of reg:squarederror.