Controle de Outliers

import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
import lasio
%matplotlib inline

print("installed")

Input_path="/home/ ....  /OtwayBasin/*.las"
import glob
my_list=[glob.glob(Input_path)]
print("Número de poços carregados",len(my_list[0])) #>>> para exibir o numero de pocos carregados

#Todos os pocos carregados com pseudonome "P"
for i in range(len(my_list[0])):
    globals()["P" + str(i)]=lasio.read(my_list[0][i]).df()
    globals()["P" + str(i)]= globals()["P" + str(i)].reindex(sorted(globals()["P" + str(i)].columns), axis=1)
    i=i+1    

#vamos imputar os dados faltantes TESTE
lista_colunas=['CALI','DRHO','DT','GR','NPHI','RDEP','RHOB','RMED','RMIC','SP','WELL']
print(len(lista_colunas))
print(len(globals()["P" + str(1)].columns))
print(list(set(lista_colunas) -set(P1.columns)))
P1.columns

for i in range(len(my_list[0])):
    if(list(set(lista_colunas) -set(globals()["P" + str(i)].columns)) != []):
        diff_list = list(set(lista_colunas) -set(globals()["P" + str(i)].columns))
        print(i,diff_list)
        for j in range(len(diff_list)): #adiciona cada coluna por vez de acordo c a diff_list
            globals()["P" + str(i)][diff_list[j]] = pd.Series(np.nan, index=globals()["P" + str(i)].index)
            
        j=j+1
    #reordena por ordem alfabética cada dataframe
    globals()["P" + str(i)]= globals()["P" + str(i)].reindex(sorted(globals()["P" + str(i)].columns), axis=1)    
    i=i+1
    print(i," ____")
    

P29

#todos os pocos com os perfis


lista_pocos=['P1','P9','P10','P14','P17','P19','P25','P26','P30','P35','P37','P38','P47','P49','P53','P58','P62','P64','P65','P70','P74','P77']

all_wells=pd.DataFrame()
all_wells=P1
#for i in range(len(my_list[0])):
for i in range(len(lista_pocos)):
    all_wells=all_wells.append((globals()[str(lista_pocos[i])]))
    i=i+1

all_wells.describe()

#Counting all Nan values(percentage) in each well log
Tot_nan=all_wells.isna().sum()
Tot_not_nan=all_wells.count()#essa variável nao leva em consideração numero de nan

100*Tot_nan/(Tot_nan+Tot_not_nan)
#in percentage : missing values

all_wells_D=all_wells.drop(['CALI','RDEP','RMED','RMIC','SP','DRHO','PEF','NEUT','MINV','MNOR','WELL'],axis=1)

all_wells_dropped = all_wells_D.dropna(axis=0, how='any')
print("ANTES DO DROPNA",all_wells.shape)
print("DEPOIS DO DROPNA",all_wells_dropped.shape)
print("  ")
A=(145975*11/306849*15)
print("REDUÇÃO DO DADO EM %.3f"% A)

all_wells_dropped.head()

import seaborn as sns

all_wells_dropped.describe()

df1=pd.DataFrame({'VALUE':all_wells_dropped['DT']})
df2=pd.DataFrame({'VALUE':all_wells_dropped['GR']})
df3=pd.DataFrame({'VALUE':all_wells_dropped['NPHI']})
df4=pd.DataFrame({'VALUE':all_wells_dropped['RHOB']})
df1['LOG'] = 'DT'
df2['LOG'] = 'GR'
df3['LOG'] = 'NPHI'
df4['LOG'] = 'RHOB'

graph=sns.boxplot(x=df1['LOG'], y=df1['VALUE'],color='orange',width=0.4)

graph.axhline(140,linewidth=1, linestyle='--',color='r')

plt.xlabel("Perfil", fontsize= 12)
plt.ylabel("Value", fontsize= 12)


plt.title("Controle de Outliers DT", fontsize= 15)
sns.set(rc={'figure.figsize':(5,6)})

plt.savefig('boxplot_dtp_com_outliers.png', transparent = True)

graph=sns.boxplot(x=df2['LOG'], y=df2['VALUE'], color='green',width=0.4)

graph.axhline(200,linewidth=1, linestyle='--',color='r')

plt.xlabel("Perfil", fontsize= 12)
plt.ylabel("Value", fontsize= 12)

plt.title("Controle de Outliers GR", fontsize= 15)

sns.set(rc={'figure.figsize':(5,6)})

plt.savefig('boxplot_gr_com_outliers.png', transparent = True)

graph=sns.boxplot(x=df3['LOG'], y=df3['VALUE'], color='purple',width=0.4)
graph.axhline(.60,linewidth=1, linestyle='--',color='r')

plt.xlabel("Perfil", fontsize= 12)
plt.ylabel("Value em %", fontsize= 12)

plt.title("Controle de Outliers NPHI", fontsize= 15)

sns.set(rc={'figure.figsize':(5,6)})

plt.savefig('boxplot_nphi_com_outliers.png', transparent = True)

graph=sns.boxplot(x=df4['LOG'], y=df4['VALUE'], color='yellow',width=0.3)

graph.axhline(3,linewidth=1, linestyle='--',color='r')
graph.axhline(1,linewidth=1, linestyle='--',color='r')

plt.xlabel("Perfil", fontsize= 12)
plt.ylabel("Value", fontsize= 12)
sns.set(rc={'figure.figsize':(5,6)})

plt.title("Controle de Outliers RHOB", fontsize= 15)
plt.savefig('boxplot_rhob_com_outliers.png', transparent = True)