ADS Capstone Chronicles Revised

17

Figure 10 Value Counts of Binary Features

4.2.1.2 Analysis Relationships between features of the patient data frame are not vital to the success of this data modeling. However, the team feels it is important to consider correlations and relationships to further assist in feature selection for the final food recommendation system. Figure 11 presents the correlation coefficients between numeric and ordinal features of the patient data frame from the Pandas .corr() function. The strongest relationship presented among the numeric and ordinal features is that of Education level and Income level, as Multivariate

expected. There are not any notoriously strong relationships that would cause the concern of multicollinearity within the food recommendation inputs. Based solely on correlation coefficients, each of these features would be plausible to include in the final data frame.

219