I am trying to determine the growth rate of a set of tumors. Running my model on the whole data set gives me an r2 of 0.80. If I split the set of tumors into a "slow-growing" group and a "fast-growing" group and run the analysis, I get a greater r2 value. Obviously if I keep dividing the data into smaller groups based on their growth rate, the r2 value of the resultant model will keep increasing. Is there an established way to determine subgroups in an analysis?