Question

if we have a sample size after removing the outliers we predicted amodel by 70% of the data and 30% of the data for R² ,the question that i want to ask when we want to make statistical analeses for verification of the data (we use the total sample after removing the outliers in colmogorove smirnov z test , error analeses, normality plot (frequency with regression standardized residual),normal p-p plot or we use 70% of the data for previous analeses steps

Answer 1

To build up the model or doing any statistical analysis we split up the whole data in two parts viz training and validation part. Training part have 70% of the data and validation part have 30% of the data. We use to build model base on training data and validate it that is cheak how much our model fits good based on validation data. We do not use the whole sample for the statistical analysis or the measures mentioned in question, it is done in training part of the data and how good the model is or how good it predicts is done by validation data. But if the sample size is less than 500 we do not split the data, we then take the whole data for analysis steps.