mi.js

  /**********************************************************************/
  /* count_: the number of occurrences of the feature (n-gram or word)  */
  /* all_count: training set size                                       */
  /* f_count_: the number of occurrences of `feature & female`          */
  /* m_count_: the number of occurrences of `feature & male`            */
  /* all_count_f: number of occurrences of `female`                     */
  /* all_count_m: number of occurrences of `male`                       */   
  /**********************************************************************/

  /* p(feature) */
  double p_ = (double)count_ / all_count;

  /* p(feature & gender) */
  double p_feature_f = (double)f_count_ / all_count;
  double p_feature_m = (double)m_count_ / all_count;

  /* mutual information */
  mi_ = p_feature_f * log(max(p_feature_f, 0.001) / (((double)all_count_f / all_count) * p_)) +
        p_feature_m * log(max(p_feature_m, 0.001) / (((double)all_count_m / all_count) * p_));

  /* p(gender|feature) */
  p_f_c_ = p_feature_f / p_;
  p_m_c_ = p_feature_m / p_;