Question about within transformation in section 4.3

[fengxiaoruo]

Hi, I am Sharooo, a master student major in Economics.

The research I am doing now requires econometric analysis remotely on the Daas platform via pyspark. So, your work really helps to me. I have some questions about the code for section 4.3.

In the notebook for codes in section 4.3, the calculation for within-group data transformation is different from that shown in the paper. It seems that The mean within each individual is subtracted from the original data, and the mean within the sample is subtracted. Why do I need to subtract the mean in sample again(As shown in the code below)?

df_train_within = spark.sql("""SELECT id, time, target + (select avg(target) from df) as target, x1 + (select avg(x1) from df) as x1, x2 + (select avg(x2) from df) as x2, x3 + (select avg(x3) from df) as x3, x4 + (select avg(x4) from df) as x4, x5 + (select avg(x5) from df) as x5, x6 + (select avg(x6) from df) as x6, x7 + (select avg(x7) from df) as x7 FROM df""") df_train_within.createOrReplaceTempView("df_train_within")