How Normalize Data Mining Min Max from Mysql in Python

问题

This is example of my data in mysql, I use lib flashext.mysql and python 3

RT      NK    NB    SU    SK    P    TNI IK   IB     TARGET
84876   902  1192  2098  3623  169   39  133  1063   94095
79194   902  1050  2109  3606  153   39  133   806   87992
75836   902  1060  1905  3166  161   39  133   785   83987
75571   902   112  1878  3190  158   39  133   635   82618
83797  1156   134  1900  3518  218   39  133   709   91604
91648  1291   127  2225  3596  249   39  133   659   99967

The formula MinMax is

(data-min)/(max-min)*0.8+0.1

I got the code normalize data from csv

import pandas as pd
df = pd.read_csv("dataset.csv")
norm = (df - df.min()) / (df.max() - df.min() )*0.8 + 0.1

I know how to count it like this

(first data of RT - min column RT data) / (max column RT- min column RT) * 0.8 + 0.1

So does the next column

(first data of NK - min column NK data) / (max column NK- min column NK) * 0.8 + 0.1

Please help me, How to normalize data from database, it call "dataset" and normalize it and input in another table call "normalize"

回答1:

Here is a SQL query that should get you started (assuming you want to calculate it per column):

   create table normalize as
     select
       (RT - min(RT)over()) / (max(RT)over() - min(RT)over()) * 0.8 + 0.1 as RT_norm
     from test;

I tested this query in sqlite3, not MySQL. It isn't necessarily optimal, but intuitively follows the formula. Notice, the over turns the min / max aggregate functions into window functions, which means they look at whole column, but the result is repeated on each row.

Todo

You would still need to:

• send the MySQl query via Python
• repeat the same code for each column
• give each column a name
• assign the resulting table to a schema (most likely)
• handle divide by 0 in case a column max and min are equal

来源：https://stackoverflow.com/questions/55084336/how-normalize-data-mining-min-max-from-mysql-in-python