ML in Geofisika

import pandas as pd
import numpy as np

# Load well log data
df = pd.read_csv('well_log.csv')

# Descriptive statistics
print(df.describe())

# Central tendency
print(f"Mean GR: {df['GR'].mean():.2f}")
print(f"Median RHOB: {df['RHOB'].median():.2f}")
print(f"Mode LITHOLOGY: {df['LITHOLOGY'].mode()[0]}")

# Spread
print(f"Std Dev GR: {df['GR'].std():.2f}")
print(f"Variance NPHI: {df['NPHI'].var():.4f}")
print(f"IQR RHOB: {df['RHOB'].quantile(0.75) - df['RHOB'].quantile(0.25):.2f}")

# Shape
print(f"Skewness GR: {df['GR'].skew():.2f}")
print(f"Kurtosis RHOB: {df['RHOB'].kurtosis():.2f}")

import matplotlib.pyplot as plt
import seaborn as sns

# Set style
sns.set_style("whitegrid")
plt.rcParams['figure.figsize'] = (12, 8)

# 1. Histogram
plt.subplot(2, 3, 1)
plt.hist(df['GR'], bins=50, edgecolor='black', alpha=0.7)
plt.xlabel('Gamma Ray (API)')
plt.ylabel('Frequency')
plt.title('GR Distribution')

# 2. Box Plot
plt.subplot(2, 3, 2)
df[['GR', 'RHOB', 'NPHI']].boxplot()
plt.ylabel('Values')
plt.title('Box Plot Comparison')

# 3. Scatter Plot (Crossplot)
plt.subplot(2, 3, 3)
plt.scatter(df['NPHI'], df['RHOB'], alpha=0.5, c=df['PHIT'], cmap='viridis')
plt.xlabel('Neutron Porosity')
plt.ylabel('Bulk Density')
plt.colorbar(label='Total Porosity')
plt.title('NPHI vs RHOB Crossplot')

# 4. Correlation Heatmap
plt.subplot(2, 3, 4)
corr_matrix = df[['GR', 'RHOB', 'NPHI', 'DT', 'PHIT']].corr()
sns.heatmap(corr_matrix, annot=True, fmt='.2f', cmap='coolwarm', center=0)
plt.title('Correlation Matrix')

# 5. Depth Plot (Log Plot)
plt.subplot(2, 3, 5)
plt.plot(df['GR'], df['DEPTH'])
plt.gca().invert_yaxis()
plt.xlabel('GR (API)')
plt.ylabel('Depth (m)')
plt.title('Gamma Ray Log')

# 6. Pair Plot (using seaborn)
# sns.pairplot(df[['GR', 'RHOB', 'NPHI', 'LITHOLOGY']], hue='LITHOLOGY')

plt.tight_layout()
plt.show()

from scipy.stats import pearsonr, spearmanr

# Pearson correlation (linear)
corr_pearson, p_value = pearsonr(df['NPHI'], df['PHIT'])
print(f"Pearson correlation NPHI-PHIT: {corr_pearson:.3f} (p={p_value:.4f})")

# Spearman correlation (monotonic)
corr_spearman, p_value = spearmanr(df['GR'], df['VCLAY'])
print(f"Spearman correlation GR-VCLAY: {corr_spearman:.3f} (p={p_value:.4f})")

# Full correlation matrix
features = ['GR', 'RHOB', 'NPHI', 'DT', 'RESDEEP']
corr_matrix = df[features].corr()

# Find highly correlated pairs (>0.7)
high_corr = []
for i in range(len(corr_matrix.columns)):
    for j in range(i+1, len(corr_matrix.columns)):
        if abs(corr_matrix.iloc[i, j]) > 0.7:
            high_corr.append((corr_matrix.columns[i], 
                            corr_matrix.columns[j], 
                            corr_matrix.iloc[i, j]))

print("\nHighly correlated pairs:")
for feat1, feat2, corr in high_corr:
    print(f"{feat1} - {feat2}: {corr:.3f}")