Figure 2

Performance of the models at each step of the analyses. Model performance with each input addition over the course of the analysis for (a) oestrogen receptor (ER) status – 6 optimal transcripts. After the addition of the six optimal microRNA transcripts, the accuracy of the model has reached 100% and there is no further improvement in the error. At this point the model is considered to contain the transcripts that most accurately model the data. Columns represent median model accuracy; lines represent mean squared error for the predictions at each step. (b) progesterone receptor (PR) status – four optimal transcripts, and (c) v-erb-b2 erythroblastic leukaemia viral oncogene homolog 2 receptor (HER2/neu) status – five optimal transcripts.