Leverage provided examples of the Splunk App for Data Science and Deep Learning

The Splunk App for Data Science and Deep Learning (DSDL) ships with more than thirty data science, deep learning, and machine learning example techniques that showcase different algorithms for classification, regression, forecasting, clustering, natural language processing (NLP), graph analytics, and data mining applied to sample data.. These example techniques are available from the Examples tab of the main menu, organized by algorithm type.

Every example includes a related Jupyter Notebook that defines how the technique is implemented. You can explore these examples and leverage the SPL code and Notebook content as a means to implement your own use-cases in DSDL.

Classifier examples

• Neural Network Classifier Example: Shows how to use a binary neural network classifier build on keras and TensorFlow.
• Logistic Regression Classifier Example: Shows a simple logistic regression using PyTorch.
• Multiclass Neural Network Classifier: Shows a simple multiclass neural network classifier using PyTorch with GPU.
• Neural Network Classifier DGA: Shows a simple neural network example using PyTorch for building a multiclass classifier applied to the DGA dataset.
• Spark Gradient Boosting Classifier DGA: Shows a simple gradient boosting model with Spark's MLLib applied to the DGA dataset.
• Explainable Machine Learning with XGBoost and SHAP: Shows how to introduce explainability in machine learning models with the help of SHAP.
• Autosklearn Classification Example: Shows an automated machine learning approach to generate a classifier with autosklearn.

Regression examples

• Example Linear Regression: Shows a simple linear regression using the TensorFlow estimator class.
• Deep Neural Network Regressor: Shows a simple regression using the TensorFlow Deep Neural Network (DNN) estimator class.
• Example XGBoost Regression: Shows a simple regression example with XGBoost.
• Example GridSearch SVM: Shows how grid search can be used with a Support Vector Regressor.
• Multivariate LSTM Regressor: Shows a multivariate Long Short-Term Memory (LSTM) network to predict AC power on an example dataset from the Splunk Machine Learning Toolkit (MLTK).

Forecasting examples

• Internet Traffic Forecast using a Convolutional Neural Network: Shows an example for forecasting a univariate time series with a convolutional neural network using TensorFlow.
• App Expense Forecast using LSTM: Shows forecasting a univariate time series with a long short term neural network using TensorFlow.
• Example Forecast with Prophet: Shows how to use the Prophet library for forecasting.

Clustering examples

• Autoencoder: Shows a basic auto encoder using TensorFlow returning hidden layer representation and reconstruction loss measurements.
• Distributed KMeans algorithm with Dask: Shows how to distribute algorithm execution with Dask using KMeans.
• Clustering with UMPA and DSCAN: Shows the dimensionality reduction technique UMAP in combination with DBSCAN for distance based clustering.
• Host Clustering using UMAP on JA3 Signatures: Demonstrates use of the JA3 encoder notebook which uses UMAP to identify similarities and differences between JA3 signatures.

NLP examples

• Entity Recognition and Extraction Example using the spaCy Library: Shows a simple NER (Named Entity Recognition) using spaCy.
• Entity Recognition and Extraction Example for Japanese using spaCy + Ginza Library: Shows a simple NER (Named Entity Recognition) for Japan ese using spaCy and Ginza.
• Sentiment Analysis using spaCy: Shows a simple sentiment analysis using spaCy.

Graphs examples

• Graph Analysis Example for Bitcoin Transactions: Shows how to calculate centrality measures in a bitcoin transaction graph using NetworkX.
• Graph Analysis Example for Community Detection with Louvain Modularity: Shows the Louvain Modularity method for community detection running on GPU.
• Causal Inference: Shows how you can use Bayesian Networks to combine machine learning and domain expertise for causal reasoning.

Data Mining examples

• Frequent Itemsets for Shopping Analysis: Shows how to find frequent itemsets using FP Growth algorithm from Spark MLLib.
• Collaborative Filtering Recommendations: Shows how to get recommendations using Collaborative Filtering from Spark MLLiib.
• Rapids UMAP on DGA: Shows how to analyze the DGA dataset with Rapiid's UMAP running on GPU.
• Example for Process Mining with PM4Py: Shows how process mining with PMPy can be integrated with the app.
• Time Series Anomalies with STUMPY: Shows how to detect anomalies in time series using matrix profiles.
• Example for Bayesian Online Change Point Detection: Shows how to detect change points or drift in time series using Bayesian Online Change Point Detection.
• Anomaly Detection with Random Cut Forest: Shows how to detect anomalies using the Random Cut Forest algorithm.
• Seasonality and Trend Decomposition (STL): Shows how to decompose a time series into its trend and seasonality components.
• Example for Hidden Markov Models applied to punct notations: Shows a very simple application of creating a Hidden Markov Model (HMM) based on the sequence of characters in punct notations.
• Example Online Learning Anomaly Detection: Shows how to utilize an online learning anomaly detection model using the HalfSpaceTree algorithm in River.
• Anomaly Detection with PyOD: Shows an Unsupervised Outlier Detection using Empirical Cumulative Distribution functions utilizing the PyOD library.

Basic examples

• Correlation Matrix and Pair Plot: Show how to do a simple correlation matrix and embed graphic plots from seaborn.
• Spark Pi Example: Shows a basic hello world example for Spark Pi.
• Seasonality and Trend Decomposition (STL): Shows how to decompose a time series into its trend and seasonality components.