I ran a series of regressions for 400.000 ID's. I have the following long dataset (example for 1 ID)
ID variable measure value
7.301004e+18 (Intercept) Estimate 1.225463e+02
7.301004e+18 price Estimate -1.055974e+02
7.301004e+18 pricepromo Estimate 3.085680e-01
7.301004e+18 feature Estimate 1.629105e+00
7.301004e+18 display Estimate 2.171643e+01
7.301004e+18 trend Estimate -1.148725e-02
7.301004e+18 addition_step Estimate -4.813033e-01
7.301004e+18 (Intercept) Std. Error 1.674007e+01
7.301004e+18 price Std. Error 1.724551e+01
7.301004e+18 pricepromo Std. Error 2.051796e-01
7.301004e+18 feature Std. Error 3.010596e+00
7.301004e+18 display Std. Error 3.580683e+00
7.301004e+18 trend Std. Error 1.297774e-02
7.301004e+18 addition_step Std. Error 2.400008e+00
7.301004e+18 (Intercept) Pr(>|t|) 1.022462e-12
7.301004e+18 price Pr(>|t|) 1.885259e-09
7.301004e+18 pricepromo Pr(>|t|) 1.332546e-01
7.301004e+18 feature Pr(>|t|) 5.886688e-01
7.301004e+18 display Pr(>|t|) 2.645076e-09
7.301004e+18 trend Pr(>|t|) 3.765107e-01
7.301004e+18 addition_step Pr(>|t|) 8.411398e-01
Which I would need to reshape wide such that I have 1 column for each combination of variable*measure. Variable has 8 levels and measure has 3 levels, so I create 21 columns.
I use the following code (based on reshape2) to do so:
reshaped <- dcast(mydata, ID ~ measure + variable, value.var = "value")
The problem is that in my output I have only 3.229 observations left (although I had 2885342 observations in the long dataset).
Also, I get the following error:
"Aggregation function missing: defaulting to length".
I do not see why I would want to aggregate. I want one observations per unique ID.
