Splunk for Analytics and Data Science (SADS)

Module 1 – Analytics Workflow

• Define terms related to analytics and data science
• Describe the analytics workflow
• Describe common usage scenarios
• Navigate Splunk Machine Learning Toolkit

Module 2 – Exploratory Data Analysis

• Describe the purpose of data exploration
• Identify SPL commands for data exploration
• Split data for testing and training using the sample command

Module 3 – Predict Numeric Fields with Regression

• Differentiate predictions from estimates
• Identify prediction algorithms and assumptions
• Describe the fit and apply commands
• Model numeric predictions in the MLTK and Splunk Enterprise
• Use the score command to evaluate models

Module 4 – Clean and Preprocess the Data

• Define preprocessing and describe its purpose
• Describe algorithms that preprocess data for use in models
• Use FieldSector to choose relevant fields
• Use PCA and ICA to reduce dimensionality
• Normalize data with StandardScaler and RobustScaler
• Preprocess text using Imputer, and NPR, TF-IDF, HashingVectorizer and the cluster command

Module 5 – Cluster Data

• Define Clustering
• Identify clustering methods, algorithms, and use cases
• Use Smart Clustering Assistant to cluster data
• Evaluate clusters using silhouette score
• Validate cluster coherence
• Describe clustering best practices

Module 6 – Anomaly Detection

• Define anomaly detection and outliers
• Identify anomaly detection use cases
• Use Splunk Machine Learning ToolKit Smart Outlier Assistant
• Detect anomalies using the Density Function algorithm
• Optimize anomaly detection with Local Outlier Factor
• View results with the Distribution Plot visualization

Module 7 – Estimation and Prediction

• Differentiate predictions from forecasts
• Use the Smart Forecasting Assistant
• Use the StateSpaceForecast algorithm
• Forecast multivariate data
• Account for periodicity in each time series

Module 8 – Classification

• Define key classification terms
• Use classification algorithms
• AutoPrediction
• LogisticRegression
• SVM (Support Vector Machines)
• RandomForestClassifier
• Evaluate classifier tradeoffs
• Evaluate results of multiple algorithms