Scribd Logo
Upload

Welcome to Scribd!

  • 66 views

    Sales Forecasting Methods

    Uploaded by

    Sushant Bharti
    This document discusses various statistical and machine learning techniques for sales forecasting. It notes that traditional time series models like ARIMA are effective for organization-level or division-level forecasts but more granular forecasts require sophisticated machine learning models. Examples are given of how forecasting needs have changed for video game and sportswear clients to require prediction at the individual item level using techniques like neural networks, random forests, and XGBoost that can handle limited data and non-linear trends. The document also outlines types of additional data needed and an approach using an ensemble of models tailored to different item volumes and durations.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd

    Sales Forecasting Methods

    Uploaded by

    Sushant Bharti
    66 views3 pages
    This document discusses various statistical and machine learning techniques for sales forecasting. It notes that traditional time series models like ARIMA are effective for organization-level or division-level forecasts but more granular forecasts require sophisticated machine learning models. Examples are given of how forecasting needs have changed for video game and sportswear clients to require prediction at the individual item level using techniques like neural networks, random forests, and XGBoost that can handle limited data and non-linear trends. The document also outlines types of additional data needed and an approach using an ensemble of models tailored to different item volumes and durations.

    Original Description:

    Forecasting Approaches

    Copyright

    © © All Rights Reserved

    Available Formats

    DOCX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    This document discusses various statistical and machine learning techniques for sales forecasting. It notes that traditional time series models like ARIMA are effective for organization-level or division-level forecasts but more granular forecasts require sophisticated machine learning models. Examples are given of how forecasting needs have changed for video game and sportswear clients to require prediction at the individual item level using techniques like neural networks, random forests, and XGBoost that can handle limited data and non-linear trends. The document also outlines types of additional data needed and an approach using an ensemble of models tailored to different item volumes and durations.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    66 views3 pages

    Sales Forecasting Methods

    Uploaded by

    Sushant Bharti
    This document discusses various statistical and machine learning techniques for sales forecasting. It notes that traditional time series models like ARIMA are effective for organization-level or division-level forecasts but more granular forecasts require sophisticated machine learning models. Examples are given of how forecasting needs have changed for video game and sportswear clients to require prediction at the individual item level using techniques like neural networks, random forests, and XGBoost that can handle limited data and non-linear trends. The document also outlines types of additional data needed and an approach using an ensemble of models tailored to different item volumes and durations.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    of 3

    Sales Forecasting Methods

    Statistical Techniques
    These statistical techniques include various well-known models that have formal statistical
    foundations:
     Moving Average
     Weighted Moving Average
     Least Square Regression
     Linear Approximation
     Second Degree Approximation
     Linear Smoothing
     Exponential Smoothing
     Holt Winters
     BOX & Jenkins Model
     Regression Models
     ARIMA & ARIMAX

    These methods have been implemented in different areas and they provide satisfactory results.
    However, their efficiency strongly depends of the field of application, the forecast goal (especially the
    horizon) or the user experience.

    The market is strongly impacted by numerous factors which make the sales very fluctuated. These
    factors, also called explanatory variables, are sometimes not controlled and even unknown. Some of
    them involve an increase of the purchase decision, others modify the store traffic. Hence, the difficulty
    to exactly identify them and to quantify their impact.

    The impact of these variable could be very dissimilar on sales. Indeed, some variables generate
    punctual fluctuations without significantly affecting the overall volume of sales, for instance a
    temporal price discount produces peaks of sales.

    Others impact more globally the sales such as macroeconomic data or strategy of retail. For instance,
    sales can show an unexpected decline which could be explained from these kinds of factors.

    Consequently, these methods are not easily and not efficiently implemented, they require large
    historical data sets, a complex optimization of their parameters, a certain experience of the operator,
    and they are limited to linear structure.

    Changing Trends in Forecasting


    Over the last few years have dealt with simple forecasting problems for which we can use
    very simple time-series forecasting techniques like ARIMA and ARIMAX or even linear
    regression these are forecasts which are more at an organization or for specific business
    divisions. But over the years we have seen a distinct shift in focus of all our clients to get
    forecasts at a more granular level, sometimes for even specific items. These forecasts are
    difficult to attain using simple techniques. This is where more sophisticated techniques come
    into play. These techniques are the more complex machine learning techniques which include
    neural network, RF, XG Boost etc.
    We cater to various industry domains and verticals and to explain how the clients’
    requirements have changed over the years I can think of two very distinct examples from two
    specialty domains, video gaming industry and sportswear manufacturer and retailer. Below I
    will try to explain how the business requirement for a forecast was different for both these
    clients.
    There are a lot of items here which are newly launched items having very few data points at
    a week level. Here, the traditional time-series methodology will fail because of lack of data
    points, also not all the styles will showcase similar trend and seasonality. Along with this,
    there will also be styles which have minimal sales and prediction for these styles is a major
    challenge for this client.
    Artificial Intelligence and Sales Optimization
    Some E-commerce giants such as Amazon are already using AI to pre-empt and optimize sales
    forecasting. In that case, a deep data analysis enables the machine to spotlight key
    information on customer online consumption's habits and then suggests products that will fit
    their needs. It will also recommend how to have an accurate reply for any prospect depending
    on the data collected through their emails, business’s CRM and so on. Such an AI use will both
    improve users experience and online seller’s sales. But there is more on AI’s forecasting
    capacity: it also enables machines to calculate the probability of a customer purchase, which
    will give the company short-term and middle-term views on First turnover.
    For a special group of items in a German super market neural nets have been trained to
    forecast future demands on the basis of the past data augmented with further influences like
    price changing advertising campaigns and holiday season information. The experimental
    results show that neural nets outperform the naive and statistical approaches that are
    currently being used in the supermarket
    Advanced Sales Forecasting Methods
     Neural Net(Deep Learning, RNN)
     XG Boost
     Random Forest

    Data needed-
     Item Descriptions
     Store Details (Size, Employees)
     Distribution Hierarchy
     Customer Details (Type, Buying pattern)
     Customer Demographics
     Promotion Details
     Discounts Details
     Return/Cancellation
     Items Association
     Reviews (Sentiment)
     Sales, Profit/Cost

    Approach
    We can develop an ensemble of models where we divided all the styles into few buckets of
     High volume – high duration
     High volume – low duration
     Low volume – high duration
     Low volume – low duration and completely new launches.
    For the styles that have high volume and high duration, we can still use a time-series or a
    regression technique but for all the others these traditional methods will have limitations.
    Hence, we needed to apply ML techniques for these styles.
    For the styles with low duration, we can use Random forest and XGB methods to arrive at the
    predictions. Also for these styles what was more important was to get a proper demand
    prediction rather than identify the drivers of sales and their impact, hence ML techniques.

    You might also like