Repackaging into TensorHouse

ikatsov · Feb 24, 2020 · 8d7fc86 · 8d7fc86
1 parent 0a7593f
commit 8d7fc86
Show file tree

Hide file tree

Showing 21 changed files with 3,654 additions and 1,171 deletions.
diff --git a/.gitignore b/.gitignore
@@ -99,3 +99,5 @@ ENV/
 
 # mypy
 .mypy_cache/
+
+.DS_Store
diff --git a/README.md b/README.md
@@ -1,3 +1,46 @@
-Models from Introduction to Algorithmic Marketing book, as well as more advanced marketng and pricing models.
+<p align="center">
+  <img src="https://github.com/ikatsov/algorithmic-marketing-examples/blob/master/resources/logo-2000x436px-gr.png" title="TensorHouse Logo">
+</p>
 
-Book's website - https://algorithmicweb.wordpress.com/
+### About
+TensorHouse is a collection of reference machine learning and optimization models for enterprise operations: marketing, pricing, supply chain, and more. The goal of the project is to provide baseline implementations for industrial, research, and educational purposes.
+
+This project contains the follwoing resources:
+* a well-documented repository of reference model implementations, 
+* a manually curated list of [important papers](https://github.com/ikatsov/tensor-house/blob/master/resources/papers.md) in modern operations research,
+* a manually curated list of [public datasets](https://github.com/ikatsov/tensor-house/blob/master/resources/datasets.md) related to entrerpirse use cases.
+
+### Illustrative Example 
+*Strategic price optimization using reinforcement learning*
+![Price Optimization Using RL Animation](https://github.com/ikatsov/algorithmic-marketing-examples/blob/master/resources/hilo-pricing-dqn-training-animation.gif)
+
+### List of Models 
+
+* Promotions and Advertisements
+   * Campaign/Channel Attribution using Adstock Model
+   * Customer Lifetime Value (LTV) Modeling using Markov Chain
+   * Next Best Action Model using Reinforcement Learning (Fitted Q Iteration)
+   * Multi-touch Multi-channel Attribution Model using Deep Learning (LSTM with Attention)
+* Search
+   * Latent Semantic Analysis (LSA)
+* Recommendations
+   * Nearest Neighbor User-based Collaborative Filtering
+   * Nearest Neighbor Item-based Collaborative Filtering
+   * Item2Vec Model using NLP Methods (word2vec)
+   * Customer2Vec Model using NLP Methods (doc2vec)
+* Pricing and Assortment
+  * Markdown Price Optimization
+  * Dynamic Pricing using Thompson Sampling
+  * Dynamic Pricing with Limited Price Experimentation
+  * Price Optimization using Reinforcement Learning (DQN)
+* Supply Chain
+  * Multi-echelon Inventory Optimization using Reinforcement Learning (DDPG, TD3)
+
+### Approach
+* The most basic models come from Introduction to Algorithmic Marketing book. Book's website - https://algorithmicweb.wordpress.com/
+* More advanced models use deep learning techniques to analyze event sequences (e.g. clickstream) and reinforcement learning for optimization (e.g. safety stock management policy)
+* Almost all models are based on industrial reports and real-life case studies
+
+### Contributors
+* Ilya Katsov
+* Dmytro Zikrach 
diff --git a/pricing/dynamic-pricing-limited-experimentation.ipynb b/pricing/dynamic-pricing-limited-experimentation.ipynb
@@ -1,28 +1,32 @@
 {
  "cells": [
   {
-   "cell_type": "code",
-   "execution_count": null,
+   "cell_type": "markdown",
    "metadata": {
-    "collapsed": true
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
-   "outputs": [],
    "source": [
-    "# This is an implementation of the dynamic pricing model from\n",
-    "#\n",
-    "# Cheung, Wang Chi and Simchi-Levi, David and Wang, He, \n",
-    "# Dynamic Pricing and Demand Learning with Limited Price Experimentation \n",
-    "# (February 26, 2017). Operations Research, Forthcoming. \n",
-    "# Available at SSRN: https://ssrn.com/abstract=2457296 or http://dx.doi.org/10.2139/ssrn.2457296\n",
-    "#\n",
-    "# See also https://blog.griddynamics.com/dynamic-pricing-algorithms/"
+    "# Dynamic Pricing Model Using Hypothesis Testing\n",
+    "\n",
+    "We implement a dynamic price optimization model that generates a set of hypothesis (using historical data or some alternative approach) and optimizes price through ascent in this hypothesis space.\n",
+    "\n",
+    "| Property | Value |\n",
+    "|--|--|\n",
+    "| Description | D004 (see [descriptions](https://github.com/ikatsov/tensor-house/blob/master/resources/descriptions.md)) |\n",
+    "| Dataset | NA |\n",
+    "| Papers | Cheung2017 (see [papers](https://github.com/ikatsov/tensor-house/blob/master/resources/papers.md)) | "
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 2,
    "metadata": {
-    "collapsed": true
+    "collapsed": true,
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
    "outputs": [],
    "source": [
@@ -137,7 +141,10 @@
    "cell_type": "code",
    "execution_count": 12,
    "metadata": {
-    "collapsed": true
+    "collapsed": true,
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
    "outputs": [],
    "source": [
@@ -235,7 +242,10 @@
    "cell_type": "code",
    "execution_count": 14,
    "metadata": {
-    "collapsed": true
+    "collapsed": true,
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
    "outputs": [],
    "source": [
@@ -274,9 +284,9 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.8"
+   "version": "3.7.4"
   }
  },
  "nbformat": 4,
- "nbformat_minor": 2
+ "nbformat_minor": 4
 }
diff --git a/pricing/dynamic-pricing-thompson.ipynb b/pricing/dynamic-pricing-thompson.ipynb
@@ -1,29 +1,28 @@
 {
  "cells": [
   {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {
-    "collapsed": true
-   },
-   "outputs": [],
+   "cell_type": "markdown",
+   "metadata": {},
    "source": [
-    "# This is an implementation of the dynamic pricing model from\n",
-    "#\n",
-    "# Ferreira, Kris and Simchi-Levi, David and Wang, He, \n",
-    "# Online Network Revenue Management Using Thompson Sampling (November 7, 2017). \n",
-    "# Operations Research, Forthcoming. \n",
-    "# Available at SSRN: https://ssrn.com/abstract=2588730 or http://dx.doi.org/10.2139/ssrn.2588730\n",
-    "# https://pdfs.semanticscholar.org/6dce/31c44fb1ad1b8dbe231f7654c37506daa909.pdf\n",
+    "# Dynamic Pricing Model Using Thompson Sampling\n",
+    "\n",
+    "We implement a dynamic price optimization model for descrete price levels using Thompson sampling.\n",
     "\n",
-    "# See also https://blog.griddynamics.com/dynamic-pricing-algorithms/"
+    "| Property | Value |\n",
+    "|--|--|\n",
+    "| Description | D004 (see [descriptions](https://github.com/ikatsov/tensor-house/blob/master/resources/descriptions.md)) |\n",
+    "| Dataset | NA |\n",
+    "| Papers | Ferreira2017 (see [papers](https://github.com/ikatsov/tensor-house/blob/master/resources/papers.md)) | "
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 16,
    "metadata": {
-    "collapsed": true
+    "collapsed": true,
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
    "outputs": [],
    "source": [
@@ -46,7 +45,10 @@
    "cell_type": "code",
    "execution_count": 17,
    "metadata": {
-    "collapsed": true
+    "collapsed": true,
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
    "outputs": [],
    "source": [
@@ -210,7 +212,10 @@
    "cell_type": "code",
    "execution_count": 23,
    "metadata": {
-    "collapsed": true
+    "collapsed": true,
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
    "outputs": [],
    "source": [
@@ -251,9 +256,9 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.8"
+   "version": "3.7.4"
   }
  },
  "nbformat": 4,
- "nbformat_minor": 2
+ "nbformat_minor": 4
 }
diff --git a/pricing/markdown-optimization.ipynb b/pricing/markdown-optimization.ipynb
@@ -1,27 +1,34 @@
 {
  "cells": [
   {
-   "cell_type": "code",
-   "execution_count": 141,
+   "cell_type": "markdown",
    "metadata": {
-    "collapsed": true
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
-   "outputs": [],
    "source": [
-    "# Models from Introduction to Algorithmic Marketing\n",
-    "# https://algorithmicweb.wordpress.com/\n",
-    "#\n",
-    "# This model optimizes prices for N consecutive time periods \n",
-    "# each of which has its own demand function. It is assumed that \n",
-    "# the stock level of the product is limited and the goal\n",
-    "# is to sell out the stock in time maximizing the revenue."
+    "# Basic Markdown Optimization Model\n",
+    "\n",
+    "This model optimizes prices for N consecutive time periods \n",
+    "each of which has its own demand function. It is assumed that \n",
+    "the stock level of the product is limited and the goal\n",
+    "is to sell out the stock in time maximizing the revenue.\n",
+    "\n",
+    "| Property | Value |\n",
+    "|--|--|\n",
+    "| Description | See [Introduction to Algorithmic Marketing](https://algorithmicweb.wordpress.com/ ) book |\n",
+    "| Dataset | Synthetic |"
    ]
   },
   {
    "cell_type": "code",
    "execution_count": 142,
    "metadata": {
-    "collapsed": true
+    "collapsed": true,
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
    "outputs": [],
    "source": [
@@ -41,7 +48,10 @@
    "cell_type": "code",
    "execution_count": 143,
    "metadata": {
-    "collapsed": true
+    "collapsed": true,
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
    "outputs": [],
    "source": [
@@ -106,7 +116,10 @@
    "cell_type": "code",
    "execution_count": 145,
    "metadata": {
-    "collapsed": true
+    "collapsed": true,
+    "jupyter": {
+     "outputs_hidden": true
+    }
    },
    "outputs": [],
    "source": [
@@ -214,9 +227,9 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.8"
+   "version": "3.7.4"
   }
  },
  "nbformat": 4,
- "nbformat_minor": 2
+ "nbformat_minor": 4
 }