python-数据处理-合并表

# 1读取数据
#.2.合并表
#3.找到user_id和aisle_之间的关系
import pandas as pd

#1.获取数据
order_products=pd.read_csv("C:/Users/18578/Desktop/instacart/order_products__prior.csv")
products=pd.read_csv('C:/Users/18578/Desktop/instacart/products.csv')
orders=pd.read_csv('C:/Users/18578/Desktop/instacart/orders.csv')
aisles=pd.read_csv('C:/Users/18578/Desktop/instacart/aisles.csv')

tab1=pd.merge(aisles,products,on=["aisle_id","aisle_id"])

tab2=pd.merge(tab1,order_products,on=['product_id','product_id'])

tab3=pd.merge(tab2,orders,on=["order_id","order_id"])

tab3

	aisle_id	aisle	product_id	product_name	department_id	order_id	add_to_cart_order	reordered	user_id	eval_set	order_number	order_dow	order_hour_of_day	days_since_prior_order
0	1	prepared soups salads	209	Italian Pasta Salad	20	94246	5	0	114082	prior	26	0	20	1.0
1	1	prepared soups salads	22853	Pesto Pasta Salad	20	94246	4	0	114082	prior	26	0	20	1.0
2	4	instant foods	12087	Chicken Flavor Ramen Noodle Soup	9	94246	15	0	114082	prior	26	0	20	1.0
3	4	instant foods	47570	Original Flavor Macaroni & Cheese Dinner	9	94246	14	1	114082	prior	26	0	20	1.0
4	13	prepared meals	10089	Dolmas	20	94246	25	0	114082	prior	26	0	20	1.0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
32434484	134	specialty wines champagnes	47713	Sparkling Rose	5	3014872	1	0	63218	prior	1	1	14	NaN
32434485	134	specialty wines champagnes	49562	Blanc De Noirs Sparkling Wine	5	34570	1	1	37901	prior	13	3	9	13.0
32434486	134	specialty wines champagnes	49562	Blanc De Noirs Sparkling Wine	5	250923	1	1	26431	prior	27	5	11	10.0
32434487	134	specialty wines champagnes	49562	Blanc De Noirs Sparkling Wine	5	1319402	1	1	26431	prior	34	2	14	7.0
32434488	134	specialty wines champagnes	49562	Blanc De Noirs Sparkling Wine	5	2298986	1	0	37901	prior	9	2	11	13.0

32434489 rows × 14 columns

table=pd.crosstab(tab3['user_id'],tab3['aisle'])

data=table[:1000]

#PCSA 降维
from sklearn.decomposition import PCA

# 1) 实例化一个转换器类
transfer=PCA(n_components=0.95)
#2) 调用fit——transform 
data_new=transfer.fit_transform(data)

data_new.shape

(1000, 35)