Laferriere Transform

laferriere.py

#Laferriere Transform
print("scatterplot of transformed values w Z scores")

def laferriere_transform (data, method):
    
    dataFrame = pd.DataFrame()

    if(method=='mean'):
        df_scaled = pd.DataFrame(StandardScaler().fit_transform(data),columns=data.columns).set_index(data.index)
        for c in data.columns:
            lower_index_ = df_scaled[[c]].sort_values(kind="quicksort", by=c,ascending=False)[df_scaled[[c]].sort_values(kind="quicksort", by=c,ascending=False)<0].cumsum().dropna()
            lower_index_ = lower_index_/lower_index_.min()
            lower_index_ = abs(((lower_index_-.5)-.5))/2
            upper_index_ = df_scaled[[c]].sort_values(kind="quicksort", by=c,ascending=True)[df_scaled[[c]].sort_values(kind="quicksort", by=c,ascending=True)>=0].cumsum().dropna()            
            upper_index_ = upper_index_/upper_index_.max()/2+.5            
            together_index = pd.concat([lower_index_,upper_index_],axis=0)            
            together_index.columns = [c]
            together_index = together_index.sort_values(kind="quicksort", by=c,ascending=True)
            dataFrame = pd.concat([dataFrame,together_index],axis=1)            
                    
            plt.scatter(pd.concat([together_index,df_scaled[[c]]],axis=1).iloc[:,0],pd.concat([together_index,df_scaled[[c]]],axis=1).iloc[:,1])
            
           
    elif(method=='median'):
        df_scaled = pd.DataFrame((data-np.median(data,axis=0))/stats.median_abs_deviation(data),columns=data.columns).set_index(data.index)
        for c in data.columns:
            lower_index_ = df_scaled[[c]].sort_values(kind="quicksort", by=c,ascending=False)[df_scaled[[c]].sort_values(kind="quicksort", by=c,ascending=False)<0].cumsum().dropna()
            lower_index_ = lower_index_/lower_index_.min()
            lower_index_ = abs(((lower_index_-.5)-.5))/2
            upper_index_ = df_scaled[[c]].sort_values(kind="quicksort", by=c,ascending=True)[df_scaled[[c]].sort_values(kind="quicksort", by=c,ascending=True)>=0].cumsum().dropna()
            upper_index_ = upper_index_/upper_index_.max()/2+.5
            together_index = pd.concat([lower_index_,upper_index_],axis=0)
            together_index.columns = [c]
            together_index = together_index.sort_values(kind="quicksort", by=c,ascending=True)
            dataFrame = pd.concat([dataFrame,together_index],axis=1)
            
            plt.scatter(pd.concat([together_index,df_scaled[[c]]],axis=1).iloc[:,0],pd.concat([together_index,df_scaled[[c]]],axis=1).iloc[:,1])
            #plt.show()

    plt.show()
    return(dataFrame)        

median