Data Science & ML Claudeスキル

Streaming Inference

Implements real-time machine learning architectures for processing unbounded data streams with sub-100ms prediction latency.

Ground Truth Labeling

Establishes rigorous and defensible ground truth labels for evaluation datasets based on authoritative guidelines.

Feature Engineering Patterns

Transforms raw data into optimized features to improve machine learning model performance and predictive accuracy.

Base Rate Analysis

Combats decision-making bias by anchoring probability assessments on statistical baseline frequencies before incorporating specific case details.

Agent-Based Modeling Simulation

Simulates complex systems from the bottom-up by defining simple rules for individual agents to observe emergent collective patterns.

Regression to the Mean Thinking Pattern

Enhances decision-making by identifying statistical reversion in performance data and preventing false causal interpretations.

Survivorship Bias Thinking Pattern

Enables Claude to identify and mitigate logical errors caused by focusing on visible successes while ignoring hidden failures.

Clustering Illusion Analyzer

Identifies and mitigates the tendency to see meaningful patterns in random data streaks or clusters.

LLM Fine-Tuning Strategies

Provides comprehensive frameworks and best practices for adapting foundation models to specialized domains using full fine-tuning and parameter-efficient methods.

Academic Research Data Analyst

Streamlines academic research data analysis by enforcing reproducible dbt pipelines and interactive Streamlit dashboards.

Academic Data Analyst

Facilitates reproducible academic research and data pipelines using dbt and Streamlit while enforcing rigorous data integrity standards.

Training Set Builder

Extracts structured training examples from document sets to create high-quality datasets for teaching LLMs specific tasks or styles.

Real-World Evidence Analysis in R

Generates real-world evidence from observational data using advanced R-based causal inference and target trial emulation techniques.

OpenAI Whisper Transcription

Transcribes audio and video files locally using the OpenAI Whisper CLI without the need for an API key.

ML Model Selection Framework

Guides the selection of optimal machine learning algorithms by analyzing problem structure, data properties, and production constraints.

Availability Heuristic Thinking Pattern

Mitigates cognitive bias in decision-making by prioritizing statistical base rates over vivid, anecdotal evidence.

Stock Analyzer

Performs professional financial stock analysis using technical indicators, automated chart generation, and news sentiment tracking.

Training Loop Patterns

Implements production-grade deep learning training loops using battle-tested architectural patterns for optimized performance and stability.

Groq & Ollama AI Bridge

Enables high-speed chat completions via Groq Cloud and local text embeddings through Ollama for efficient RAG workflows.

Autonomous Agent Architect

Architects reliable, production-ready AI agent workflows using constrained loops and proven reliability patterns.

Sensitivity Analysis

Performs systematic testing of input variables to identify key drivers and assess model risk across finance, engineering, and strategy.

Base Rate Fallacy Logic

Corrects probability judgments by integrating statistical base rates with case-specific information to avoid common cognitive biases.

Data Network Effects

Leverages the data flywheel mental model to build compounding competitive advantages through automated machine learning and user-generated signals.

Online Learning System Design

Implements continuous model updates and incremental learning patterns to handle evolving data streams without full retraining.

Supply and Demand Analysis

Predicts market price and quantity changes by analyzing the interaction between producer supply and consumer demand at equilibrium.

RAG Thinking Patterns

Implements sophisticated Retrieval-Augmented Generation architectures to ground AI responses in real-time external knowledge and eliminate hallucinations.

Gambler's Fallacy Recognition

Identifies and mitigates the gambler's fallacy to improve decision-making in probabilistic and high-risk scenarios.

Veo 3.1 Consistent Video Generation

Generates high-quality videos with consistent subject appearance using Google’s Veo 3.1 model and reference images.

Scientific Critical Thinking

Evaluates research rigor by assessing methodology, statistical validity, and evidence quality using industry-standard frameworks like GRADE and Cochrane.

Excel Data Analysis

Analyzes, manipulates, and visualizes Excel spreadsheets using Python-based automation to generate insights and formatted reports.