searching in range between columns using sqlite3 in pandas

Question

I have found solution to my problem in one question Merge pandas dataframes where one value is between two others I tried to modify it for my situation but it didn't work. In code given below, I need df to show the beginning and ending for each sold product and category. But it ignores date being between start and end period. as can be seen on screen for sel of Apple on 01/06/2020 period 26/03/2020 - 31/07/2020 should be shown but it shows other. How should I clarify SQL query?

import pandas as pd
import sqlite3

dates_of_discount=pd.DataFrame({"Date_begining":['01/01/2021','01/02/2020','26/03/2020'],
                   "Date_ending":['31/12/2021', '25/02/2020', '31/07/2020'],
                   "Category":['Discount', 'Not Discount', "Discount"],
                   "d_Product":['Apple', 'Peach', "Apple"]})
purchase_dates=pd.DataFrame({"date":(["20/01/2020", "18/02/2020", "01/06/2020"]),
                          "Qty":[100, 200, 300],
                          "Price":[3.5,4, 20],
                          "p_Product":['Apple', 'Peach', "Apple"]})


conn = sqlite3.connect(':memory:')

dates_of_discount.to_sql('disc', conn, index=False)
purchase_dates.to_sql('purch', conn, index=False)

qry = '''
    select  
        purch.date Sold,
        purch.p_Product Prod,
        purch.Qty,
        purch.Price,
        Date_begining Period_Start,
        Date_ending Period_End,
        Category Output
    from
        purch join disc on
        date between Date_begining and Date_ending and
        d_Product = p_Product
    '''
df = pd.read_sql_query(qry, conn)
df

A date format like `01/06/2020` is not comparable. Change to `YYYY-MM-DD` — forpas, May 14 '22 at 13:01
thanks! it helped, but if instead of dates there are numbers, let's say size of product. Min and Max. How to make it work? — Zaur Guliyev, May 14 '22 at 13:14

score 1 · Accepted Answer · answered May 14 '22 at 16:51

As commented, by properly converting the date strings to actual datetime in pandas using pd.to_datetime, the SQL join operation should return expected results:

Input Data (with date conversion)

dates_of_discount = pd.DataFrame({
    "Date_begining": pd.to_datetime(
        ['01/01/2021','01/02/2020','26/03/2020'],
        format="%d/%m/%Y"
    ),
    "Date_ending": pd.to_datetime(
        ['31/12/2021', '25/02/2020', '31/07/2020'],
        format="%d/%m/%Y"
    ),
    "Category": ['Discount', 'Not Discount', "Discount"],
    "d_Product": ['Apple', 'Peach', "Apple"]
})

purchase_dates=pd.DataFrame({
    "date": pd.to_datetime(
        ["20/01/2020", "18/02/2020", "01/06/2020"],
        format="%d/%m/%Y"
    ),
    "Qty":[100, 200, 300],
    "Price":[3.5,4, 20],
    "p_Product":['Apple', 'Peach', "Apple"]
})

SQLite Query

conn = sqlite3.connect(':memory:')

dates_of_discount.to_sql('disc', conn, index=False)
purchase_dates.to_sql('purch', conn, index=False)

qry = '''
    select  
        purch.date as Sold,
        purch.p_Product as Prod,
        purch.Qty,
        purch.Price,
        disc.Date_begining as Period_Start,
        disc.Date_ending as Period_End,
        disc.Category as Output
    from purch 
    join disc 
      on purch.date between disc.Date_begining and disc.Date_ending 
      and purch.p_Product = disc.d_Product
    '''
merge_df = pd.read_sql_query(qry, conn)
merge_df
#                   Sold   Prod  Qty  Price         Period_Start           Period_End        Output
# 0  2020-02-18 00:00:00  Peach  200    4.0  2020-02-01 00:00:00  2020-02-25 00:00:00  Not Discount
# 1  2020-06-01 00:00:00  Apple  300   20.0  2020-03-26 00:00:00  2020-07-31 00:00:00      Discount

By the way, pandas can also run a similar operation with merge by product and query or filter by dates (reindex and set_axis to subset and rename columns):

merge_df = (
    purchase_dates.merge(
        dates_of_discount, left_on="p_Product", right_on="d_Product"   
    ).query(
        "date >= Date_begining & date <= Date_ending"
    ).reset_index(drop=True)
    .reindex(
        ["date", "p_Product", "Qty", "Price", "Date_begining", "Date_ending", "Category"],
        axis = "columns"
    ).set_axis(
        ["Sold", "Prod", "Qty", "Price", "Period_Start", "Period_End", "Output"],
        axis = "columns",
        inplace = False
    )
)
merge_df_pd
#         Sold   Prod  Qty  Price Period_Start Period_End        Output
# 0 2020-06-01  Apple  300   20.0   2020-03-26 2020-07-31      Discount
# 1 2020-02-18  Peach  200    4.0   2020-02-01 2020-02-25  Not Discount

Finally, per your comment, same logic should work if using numbers like product size instead of dates for both SQL or pandas:

select  
    purch.date as Sold,
    purch.p_Product as Prod,
    purch.Qty,
    purch.Price,
    disc.min_product_size,
    disc.max_product_size,
    disc.Category as Output
from purch 
join disc 
  on purch.product_size between disc.min_product_size and disc.max_product_size
  and purch.p_Product = disc.d_Product

merge_df = (
    purchase_dates.merge(
        dates_of_discount, left_on="p_Product", right_on="d_Product"   
    ).query(
        "product_size >= min_product_size & product_size <= max_product_size"
    ).reset_index(drop=True)
    .reindex(
        ["date", "p_Product", "Qty", "Price", "min_product_size", "max_product_size", "Category"],
        axis = "columns"
    ).set_axis(
        ["Sold", "Prod", "Qty", "Price", "min_product_size", "max_product_size", "Output"],
        axis = "columns",
        inplace = False
    )
)

thank you very much for such a detailed explanation! it worked! — Zaur Guliyev, May 14 '22 at 18:28

searching in range between columns using sqlite3 in pandas

1 Answers1