bizlens 2.2.11

Educational descriptive analytics + statistical inference + process mining. Rich tables, diagnostic checks, hypothesis testing, and business process analysis for teaching and analysis.

BizLens v2.2.11 📊

Educational Analytics: Descriptive + Statistical Inference + Process Mining

BizLens is a Python library for business analysts, data scientists, and educators. It combines three powerful analytics domains:

• Descriptive Analytics: Rich statistical tables, summaries, and data exploration
• Statistical Inference: Hypothesis testing, confidence intervals, effect sizes
• Process Mining: Event log analysis, case metrics, bottleneck detection

---

Quick Start

Installation

pip install bizlens==2.2.11

3-Line Example: Descriptive Analytics

import bizlens as bl
import pandas as pd

data = pd.DataFrame({'revenue': [100, 150, 200, 250], 'region': ['A', 'B', 'A', 'B']})
bl.describe(data)  # Auto-detects data type, creates summary tables

Process Mining (Auto-Detection)

event_log = bl.generate_hr_onboarding_event_log(num_cases=50)
bl.describe(event_log)  # Auto-detects as event log, shows case metrics & variants

Hypothesis Testing

import numpy as np

sample = np.random.normal(100, 15, size=50)
bl.inference.confidence_interval(sample)  # 95% CI for mean
bl.inference.one_sample_ttest(sample, pop_mean=100)  # Test vs population

---

Core Modules

1. describe() — Smart Descriptive Analytics

• Analyzes DataFrames with automatic statistics
• Auto-detects event logs: When columns include case_id/activity/timestamp
• Sample vs Population: Educational comparison (ddof=1 vs ddof=0)
• Rich output: Professional console tables with rich formatting

bl.describe(dataframe)
bl.describe(event_log)  # Auto-detects and shows process mining metrics

---

2. tables — Statistical Tables

Professional tables for data summarization:

bl.tables.frequency_table(series)              # Value counts with %
bl.tables.percentile_table(series)             # Quartile breakdown (0,25,50,75,100)
bl.tables.contingency_table(df, 'row', 'col') # Crosstab with chi-square
bl.tables.summary_statistics(df)               # Count, mean, std, min, quartiles, max
bl.tables.group_comparison(df, 'group')       # ANOVA across groups
bl.tables.distribution_fit(series)             # Fit to distributions (normal, exponential, etc)
bl.tables.descriptive_comparison(df1, df2)    # Sample vs population tables

---

3. diagnostic — Data Quality & Statistical Diagnostics

Check data quality and identify anomalies:

bl.diagnostic.detect_outliers(series, method='iqr')      # IQR, Z-score, or Isolation Forest
bl.diagnostic.normality_test(series)                      # Shapiro-Wilk, Anderson-Darling, KS
bl.diagnostic.correlation_analysis(df)                    # Pearson & Spearman with heatmap
bl.diagnostic.missing_value_analysis(df)                  # Missing data patterns
bl.diagnostic.duplicate_analysis(df)                      # Find exact duplicates
bl.diagnostic.sample_vs_population(sample, pop_mean, pop_std)  # Educational t-test

---

4. inference — Hypothesis Testing & Statistical Inference

Test hypotheses and estimate population parameters:

bl.inference.confidence_interval(sample, confidence=0.95)            # 95% CI for mean
bl.inference.one_sample_ttest(sample, pop_mean=100)                 # Test sample vs population
bl.inference.two_sample_ttest(group1, group2)                       # Compare two groups (t-test + Mann-Whitney U)
bl.inference.paired_ttest(before, after)                            # Before/after testing
bl.inference.anova_test({'A': group_a, 'B': group_b, 'C': group_c}) # Multi-group comparison
bl.inference.correlation_test(x, y)                                  # Correlation significance

---

5. process_mining — Event Log Analysis

Analyze business processes from event logs:

bl.process_mining.case_metrics(event_log)           # Duration, cost, activity count per case
bl.process_mining.activity_metrics(event_log)       # Frequency, duration by activity
bl.process_mining.resource_analysis(event_log)      # Workload distribution
bl.process_mining.variant_discovery(event_log)      # Top activity sequences (paths)
bl.process_mining.bottleneck_analysis(event_log)    # Waiting time identification
bl.process_mining.rework_detection(event_log)       # Repeated activities
bl.process_mining.timeline_visualization(event_log) # Interactive Gantt chart (plotly)

Auto-Detection Requirements:

• case_id column: Uniquely identifies a case/instance
• activity column: Names the activity/step
• timestamp column: When the activity occurred
• Optional: resource, cost, or other numeric columns

---

6. quality — Data Quality Assessment

Evaluate overall data quality:

bl.quality.completeness_report(df)      # Missing data per column
bl.quality.consistency_check(df)        # Type mixing and format violations
bl.quality.uniqueness_analysis(df)      # Cardinality and duplicates
bl.quality.data_profile(df)             # Overall quality score (0-100)
bl.quality.outlier_summary(df)          # Quick outlier identification

---

Built-in Datasets & Generators

# Load classic datasets
bl.load_dataset('iris')         # Fisher's iris dataset
bl.load_dataset('tips')         # Restaurant tips
bl.load_dataset('diamonds')     # Diamond prices

# Generate synthetic business data
bl.generate_sample_data(n_rows=1000)

# Generate event logs for process mining
bl.generate_hr_onboarding_event_log(num_cases=300)
bl.generate_healthcare_event_log(num_cases=250)
bl.generate_manufacturing_event_log(num_cases=200)
bl.generate_tech_support_event_log(num_cases=400)

---

Examples

Example 1: Descriptive Analytics with Tables

import bizlens as bl
import pandas as pd
import numpy as np

# Generate data
data = pd.DataFrame({
    'region': np.random.choice(['North', 'South', 'East', 'West'], 500),
    'revenue': np.random.gamma(2, 5000, 500),
    'satisfaction': np.random.normal(7.5, 1.5, 500).clip(1, 10),
})

# Analyze
bl.describe(data)
bl.tables.frequency_table(data['region'])
bl.tables.percentile_table(data['revenue'])
bl.tables.summary_statistics(data)
bl.diagnostic.detect_outliers(data['revenue'])
bl.quality.data_profile(data)

Run this example:

python examples/01_descriptive_analytics.py

---

Example 2: Process Mining Event Logs

import bizlens as bl

# Generate HR onboarding event log
event_log = bl.generate_hr_onboarding_event_log(num_cases=100)

# Analyze with auto-detection
bl.describe(event_log)  # Auto-detects event log

# Process mining metrics
bl.process_mining.case_metrics(event_log)
bl.process_mining.variant_discovery(event_log, top_n=5)
bl.process_mining.bottleneck_analysis(event_log)
bl.process_mining.rework_detection(event_log)

Run this example:

python examples/02_process_mining_basics.py

---

Example 3: Hypothesis Testing & Inference

import bizlens as bl
import numpy as np
import pandas as pd

# Generate test data
sample = pd.Series(np.random.normal(100, 15, size=50))
control = pd.Series(np.random.normal(100, 15, size=40))
treatment = pd.Series(np.random.normal(110, 18, size=40))

# Confidence intervals
bl.inference.confidence_interval(sample, confidence=0.95)

# t-tests
bl.inference.one_sample_ttest(sample, pop_mean=100)
bl.inference.two_sample_ttest(control, treatment)

# ANOVA
bl.inference.anova_test({
    'Control': control,
    'Treatment': treatment,
})

Run this example:

python examples/03_inference_hypothesis_testing.py

---

Educational Features

Sample vs Population

Learn the difference between sample and population statistics:

# Sample: Bessel's correction (ddof=1)
bl.diagnostic.sample_vs_population(
    sample_data=my_sample,
    pop_mean=100,
    pop_std=15,
    column_name='Revenue'
)

Hypothesis Testing with Effect Sizes

Understand both statistical AND practical significance:

results = bl.inference.one_sample_ttest(sample, pop_mean=100)
print(f"p-value: {results['p_value']}")      # Statistical significance
print(f"Cohen's d: {results['cohens_d']}")   # Effect size (practical significance)

Visualization of Distributions

table, dist_info = bl.tables.distribution_fit(data['column'])
# Returns: best-fit distribution, parameters, AIC scores

---

Installation & Dependencies

Minimal Install

pip install bizlens==2.2.11

Core dependencies:

• pandas ≥1.5.0 — Data manipulation
• numpy ≥1.21.0 — Numerical operations
• scipy ≥1.9.0 — Statistical tests
• statsmodels ≥0.13.0 — Advanced statistics
• scikit-learn ≥1.0.0 — Outlier detection
• matplotlib ≥3.6.0 — Static plots
• seaborn ≥0.12.0 — Statistical plots
• rich ≥13.0.0 — Beautiful console output
• plotly ≥5.0.0 — Interactive visualizations (optional)

Optional Extras

# For Jupyter/Colab
pip install bizlens[jupyter]

# For advanced correlations
pip install bizlens[stats-advanced]

# Everything
pip install bizlens[full]

---

Running in Different Environments

Google Colab

1. Paste example code into a cell
2. Run! (Auto-install handles dependencies)

Jupyter Notebook

%run examples/01_descriptive_analytics.py

VSCode / Terminal

python examples/01_descriptive_analytics.py

Anaconda

conda install -c conda-forge bizlens==2.2.11

---

Version History

v2.2.11 (Current)

• ✅ NEW: Statistical tables module (frequency, percentile, contingency, summary)
• ✅ NEW: Diagnostic module (outliers, normality, correlations, quality checks)
• ✅ NEW: Inference module (hypothesis testing, confidence intervals, effect sizes)
• ✅ ENHANCED: Process mining with Gantt charts, bottleneck detection, rework analysis
• ✅ NEW: Quality module (data profiling, completeness, consistency assessment)
• ✅ Fixed: narwhals.selectors fallback for compatibility
• ✅ Enhanced: Rich educational docstrings with learning notes

v2.2.10

• Initial v2.2 release with descriptive analytics and process mining

v2.2.1

• Early descriptive analytics foundation

---

Future Roadmap

Version	Features	Status
v2.2.11	Tables, Diagnostic, Inference, Process Mining, Quality	✅ Current
v2.3.0	Predictive Analytics (regression, classification, decision trees)	🔄 Planned
v2.4.0	Time-Series Forecasting (ARIMA, exponential smoothing)	🔄 Planned
v2.5.0	Quality & Six Sigma (Cpk, control charts, hypothesis testing)	🔄 Planned
v3.0.0	Advanced: Deep Learning pipelines, automated ML	🔄 Future

---

Performance Notes

• Handles: DataFrames up to 1M rows (pandas) or larger (polars)
• Auto-detects: Event logs automatically (looks for case_id + activity + timestamp)
• Educational: Rich output with learning notes (not optimized for speed)
• Polars Support: Via narwhals abstraction layer for fast operations

---

FAQ

Q: Can I use BizLens with Polars DataFrames? A: Yes! Via narwhals compatibility layer. Same API works with both pandas and polars.

Q: Does BizLens replace scipy/statsmodels? A: No! It's a wrapper providing educational, easy-to-use interfaces. For advanced use, use scipy/statsmodels directly.

Q: Is it for teaching or production? A: Both! Clean API and rich output make it great for education. Lightweight and fast enough for analysis pipelines.

Q: Can I use the event log generators for my data? A: Yes! The templates include realistic HR, healthcare, manufacturing, and support data. Modify them for your use case.

Q: Does it support missing data? A: Yes! Functions automatically drop NaN values. Use pandas .fillna() or .interpolate() for imputation first.

---

Contributing

Issues, feature requests, and PRs welcome at: https://github.com/solutiongate-learn/bizlens

---

License

MIT License — Free for personal, educational, and commercial use.

---

Citation

If you use BizLens in research or publications:

@software{bizlens2024,
  title = {BizLens: Educational Analytics for Python},
  author = {Singh, Sudhanshu},
  year = {2024},
  url = {https://github.com/solutiongate-learn/bizlens}
}

---

Made with ❤️ for data analysts, teachers, and students

GitHub • PyPI • Issues