def do_corr(self):
imageIO_vardist_all = vardistplot(self.df_sigtrain, self.df_bkgtrain,
self.v_all, self.dirmlplot,
self.p_binmin, self.p_binmax)
imageIO_vardist_train = vardistplot(self.df_sigtrain, self.df_bkgtrain,
self.v_train, self.dirmlplot,
self.p_binmin, self.p_binmax)
imageIO_scatterplot = scatterplot(self.df_sigtrain, self.df_bkgtrain,
self.v_corrx, self.v_corry,
self.dirmlplot, self.p_binmin,
self.p_binmax)
imageIO_corr_sig_all = correlationmatrix(self.df_sigtrain, self.v_all,
"Signal", self.dirmlplot,
self.p_binmin, self.p_binmax)
imageIO_corr_bkg_all = correlationmatrix(self.df_bkgtrain, self.v_all,
"Background", self.dirmlplot,
self.p_binmin, self.p_binmax)
imageIO_corr_sig_train = correlationmatrix(self.df_sigtrain,
self.v_train, "Signal",
self.dirmlplot,
self.p_binmin,
self.p_binmax)
imageIO_corr_bkg_train = correlationmatrix(self.df_bkgtrain,
self.v_train, "Background",
self.dirmlplot,
self.p_binmin,
self.p_binmax)
return imageIO_vardist_all, imageIO_vardist_train, imageIO_scatterplot, \
imageIO_corr_sig_all, imageIO_corr_bkg_all, imageIO_corr_sig_train, \
imageIO_corr_bkg_train
def do_correlation(df_sig_train, df_bkg_train, var_all, var_corr_x, var_corr_y,
plotdir):
imageIO_vardist = vardistplot(df_sig_train, df_bkg_train, var_all, plotdir)
imageIO_scatterplot = scatterplot(df_sig_train, df_bkg_train, var_corr_x,
var_corr_y, plotdir)
imageIO_corr_sig = correlationmatrix(df_sig_train, plotdir, "signal")
imageIO_corr_bkg = correlationmatrix(df_bkg_train, plotdir, "background")
return imageIO_vardist, imageIO_scatterplot, imageIO_corr_sig, imageIO_corr_bkg
def do_corr(self):
imageIO_vardist = vardistplot(self.df_sigtrain, self.df_bkgtrain,
self.v_all, self.dirmlplot,
self.p_binmin, self.p_binmax)
imageIO_scatterplot = scatterplot(self.df_sigtrain, self.df_bkgtrain,
self.v_corrx, self.v_corry,
self.dirmlplot, self.p_binmin, self.p_binmax)
imageIO_corr_sig = correlationmatrix(self.df_sigtrain, self.dirmlplot,
"signal", self.p_binmin, self.p_binmax)
imageIO_corr_bkg = correlationmatrix(self.df_bkgtrain, self.dirmlplot,
"background", self.p_binmin, self.p_binmax)
return imageIO_vardist, imageIO_scatterplot, imageIO_corr_sig, imageIO_corr_bkg
def do_corr(self):
if self.step_done("distributions_correlations"):
return
self.logger.info("Make feature distributions and correlation plots")
def make_plot_name(output, label, n_var, binmin, binmax):
return f'{output}/CorrMatrix_{label}_nVar{n_var}_{binmin:.1f}_{binmax:.1f}.png'
vardistplot(self.df_sigtrain, self.df_bkgtrain, self.v_all,
self.dirmlplot, self.p_binmin, self.p_binmax,
self.p_plot_options)
if self.v_selected:
vardistplot(self.df_sigtrain, self.df_bkgtrain, self.v_selected,
self.dirmlplot, self.p_binmin, self.p_binmax,
self.p_plot_options)
vardistplot(self.df_sigtrain, self.df_bkgtrain, self.v_train,
self.dirmlplot, self.p_binmin, self.p_binmax,
self.p_plot_options)
scatterplot(self.df_sigtrain, self.df_bkgtrain, self.v_corrx,
self.v_corry, self.dirmlplot, self.p_binmin, self.p_binmax)
output = make_plot_name(self.dirmlplot, "Signal_all_vars",
len(self.v_all), self.p_binmin, self.p_binmax)
correlationmatrix(self.df_sigtrain, self.v_all, "Signal", output,
self.p_binmin, self.p_binmax, self.p_plot_options)
output = make_plot_name(self.dirmlplot, "Background_all_vars",
len(self.v_all), self.p_binmin, self.p_binmax)
correlationmatrix(self.df_bkgtrain, self.v_all, "Background", output,
self.p_binmin, self.p_binmax, self.p_plot_options)
if self.v_selected:
output = make_plot_name(self.dirmlplot, "Signal_selected_vars",
len(self.v_selected), self.p_binmin,
self.p_binmax)
correlationmatrix(self.df_sigtrain, self.v_selected, "Signal",
output, self.p_binmin, self.p_binmax,
self.p_plot_options)
output = make_plot_name(self.dirmlplot, "Background_selected_vars",
len(self.v_selected), self.p_binmin,
self.p_binmax)
correlationmatrix(self.df_bkgtrain, self.v_selected, "Background",
output, self.p_binmin, self.p_binmax,
self.p_plot_options)
output = make_plot_name(self.dirmlplot, "Signal_features",
len(self.v_train), self.p_binmin,
self.p_binmax)
correlationmatrix(self.df_sigtrain, self.v_train, "Signal", output,
self.p_binmin, self.p_binmax, self.p_plot_options)
output = make_plot_name(self.dirmlplot, "Background_features",
len(self.v_train), self.p_binmin,
self.p_binmax)
correlationmatrix(self.df_bkgtrain, self.v_train, "Background", output,
self.p_binmin, self.p_binmax, self.p_plot_options)