Ds Hypothesis Testing

Hypothesis Testing

Comprehensive guide to hypothesis testing in machine learning and data science workflows.

When to Use This Skill

• Solving real-world statistical inference problems
• Building machine learning pipelines with hypothesis testing
• Implementing best practices for hypothesis testing
• Optimizing model performance using hypothesis testing techniques
• Learning industry-standard approaches to hypothesis testing

When NOT to Use This Skill

• When using pre-built libraries without understanding underlying concepts
• For toy problems that don't require hypothesis testing rigor
• When domain expertise in specific problem requires different approach
• If your problem doesn't require the complexity this skill provides

Purpose and Key Concepts

Hypothesis Testing is a critical component of the machine learning workflow. This skill covers:

1. Theoretical foundations — Mathematical principles and statistical concepts
2. Practical implementation — Working code examples and patterns
3. Common pitfalls — Mistakes to avoid and how to recover from them
4. Best practices — Industry-standard approaches and optimization techniques

Core Workflow

1. Understand the problem — Clearly define what you're solving for
2. Select approach — Choose the right technique for your data and constraints
3. Implement solution — Write clean, tested code following best practices
4. Validate results — Verify your implementation with tests and validation
5. Optimize performance — Improve efficiency and accuracy incrementally

Implementation Patterns

Pattern 1: Basic Hypothesis Testing

import numpy as np
import pandas as pd
from scipy import stats
from typing import Dict, Any, Tuple

def perform_independent_t_test(
    group_a: np.ndarray, 
    group_b: np.ndarray, 
    alpha: float = 0.05
) -> Dict[str, Any]:
    """
    Perform an independent two-sample t-test to compare means.
    
    Args:
        group_a: Array of values for the first group
        group_b: Array of values for the second group
        alpha: Significance level for the test
        
    Returns:
        Dictionary containing t-statistic, p-value, and conclusion
    """
    if len(group_a) < 2 or len(group_b) < 2:
        raise ValueError("Each group must contain at least two observations")
        
    t_stat, p_value = stats.ttest_ind(group_a, group_b, equal_var=False)
    is_significant = p_value < alpha
    
    return {
        "t_statistic": float(t_stat)
        "p_value": float(p_value)
        "significant": bool(is_significant)
        "alpha": alpha
        "conclusion": "Reject null hypothesis" if is_significant else "Fail to reject null hypothesis"
    }

Pattern 2: Production-Ready Hypothesis Testing

import logging
import numpy as np
import pandas as pd
from scipy import stats
from typing import Any, Dict, List, Optional

logger = logging.getLogger(__name__)

class HypothesisTestingEngine:
    """Production-grade engine for statistical hypothesis testing."""
    
    def __init__(self, alpha: float = 0.05, verbose: bool = False):
        self.alpha = alpha
        self.verbose = verbose
        self.results_log: List[Dict[str, Any]] = []
        
    def run_chi_square(self, observed: List[int], expected: List[int]) -> Dict[str, Any]:
        """Execute chi-square goodness of fit test."""
        if len(observed) != len(expected):
            raise ValueError("Observed and expected arrays must have the same length")
        if any(x < 0 for x in observed) or any(x < 0 for x in expected):
            raise ValueError("Counts cannot be negative")
            
        chi2_stat, p_value = stats.chisquare(f_obs=observed, f_exp=expected)
        result = {
            "test": "chi_square"
            "statistic": float(chi2_stat)
            "p_value": float(p_value)
            "significant": bool(p_value < self.alpha)
            "timestamp": pd.Timestamp.now().isoformat()
        }
        self.results_log.append(result)
        if self.verbose:
            logger.info(f"Chi-square test completed: p={p_value:.4f}")
        return result
        
    def run_all_tests(self, data: pd.DataFrame, target_col: str, feature_col: str) -> Dict[str, Any]:
        """Run appropriate tests based on data types."""
        if target_col not in data.columns or feature_col not in data.columns:
            raise KeyError(f"Columns '{target_col}' and '{feature_col}' must exist in data")
            
        results = {}
        if data[feature_col].dtype in ['float64', 'int64']:
            groups = data.groupby(feature_col)[target_col].apply(list)
            if len(groups) >= 2:
                results['anova'] = stats.f_oneway(*groups.values())._asdict()
        else:
            contingency = pd.crosstab(data[target_col], data[feature_col])
            results['chi2'] = stats.chi2_contingency(contingency)._asdict()
            
        return results

Best Practices

• ✅ Always validate your implementation on test data
• ✅ Document your assumptions and methodology
• ✅ Use version control for reproducibility
• ✅ Monitor performance metrics in production
• ✅ Periodically review and update your approach
• ✅ Test with edge cases and outliers
• ✅ Log all significant operations for debugging

BAD vs GOOD Example

# BAD: Bypassing error handling, using magic numbers, and missing type hints
def bad_test(data):
    t, p = stats.ttest_ind(data['a'], data['b'])
    return p < 0.05  # Hardcoded threshold, no validation

# GOOD: Proper validation, type hints, configurable alpha, and clear return structure
def good_test(group_a: np.ndarray, group_b: np.ndarray, alpha: float = 0.05) -> Dict[str, Any]:
    if len(group_a) < 2 or len(group_b) < 2:
        raise ValueError("Insufficient samples for statistical testing")
    _, p_value = stats.ttest_ind(group_a, group_b, equal_var=False)
    return {
        "p_value": float(p_value)
        "significant": bool(p_value < alpha)
        "alpha_used": alpha
    }

Common Pitfalls

| Pitfall | Problem | Solution | |

Constraints

MUST DO

• Validate all data preprocessing steps are fit-only on training data, never on validation or test sets
• Implement reproducible pipelines with fixed random seeds and deterministic operations where possible
• Report model performance with confidence intervals via bootstrapping or cross-validation across multiple runs
• Log all experiments with parameters, metrics, and artifacts using MLflow or equivalent tracking system

MUST NOT DO

• Do not evaluate a model on the same data used for training — always hold out a proper test set
• Avoid overfitting to the validation set by limiting hyperparameter search iterations
• Never use features that can only be computed at inference time (look-ahead bias)
• Do not report single-run accuracy without statistical significance testing or error bars

Live References

Authoritative documentation links for this skill's domain. The model follows markdown links at load time to resolve external references and inline content.

• SciPy Stats Module
• Hypothesis Testing — Wikipedia
• t-test Documentation (SciPy)
• NIST Engineering Statistics Handbook — Hypothesis Testing
• Statistical Tests in Python (Scipy Cookbook)