|Rosella Machine Intelligence & Data Mining|
Sales trend analysis and Sales forecasting - Customer Trend Analysis
Sales trend analysis and timely identification of newly emerging customer trends is very important to businesses. Sales patterns of customer segments may indicate market trends. Upward and downward trends in sales signify new market trends. Time-series predictive modeling can be used to identify market trends embedded in changes of sales revenues. Understanding of customer and sales trends is important for marketing as well as for customer retention. Typical customer and sales trend analysis includes;
Time-Series Analysis and Projection Marketing
Time series analysis and forecasting is very important for projection marketing. Based on time-series projections, inventory levels can be maintained efficiently, for example. Retailers and wholesalers can maintain optimal inventory levels based on time-series projections. Manufacturing companies can plan manufacturing capacities. Time-series analysis and projection marketing is very important.
Segmentation methods and techniques for Customer trend analysis and Sales trend forecasting
Market Segmentation is a process that divides a market into smaller sub-markets called segments. Normally, market is segmented in such as way that customers of a segment have the same attributes. Commonly used segmentation methods and techniques for sales forecasting include the followings;
Why Customer Segmentation? People with similar attributes tend to exhibit similar purchasing patterns. This fact is particularly important in customer relationship management, marketing, and risk management. For example, people with certain life-styles tend to buy certain-types of products. Promoting products particularly targeted towards the demographic group can lead to successful marketing. Customers are segmented along the following demographic and psychographic attributes, and time-series trend analysis is performed;
PSM is a simple customer and sales trend analysis method to manage your most valuable business resources: customers and markets. Profile your customers and markets as suggested in Customer Profiling. Based on profiling, develop customer and market segments. Finally, monitor the following trends;
Sales forecasting methods and techniques: Time-series Regression
Regression is an analytic technique used in developing predictive models for numerical data. It automatically derives mathematical functions that summarize trends embedded in past historical data, in such a way that minimizes the errors between actual input data and predicted values by the models. Regression can be applied to sales time-series data. A time-series consists of a set of observations which are measured at specific time intervals, say, monthly, quarterly, yearly, etc. Observations we are interested are sales revenues.
Customer (or market) segments have different sales trends. Some segments may be growing, while others are declining. Segment-by-segment sales forecasting can produce very useful information. Forecasting can be short term, mid term and long term. Long term forecasting may not produce accurate predictions. However it is very useful in understanding market trends.
Sales forecasting with Seasonal Adjustment
Sales time-series data often contain seasonal patterns. For example, clothing and fruits sales can fluctuate based on seasons. This hides underlying sales patterns and makes it difficult to project sales figures accurately. Seasonal adjustment is used to overcome this problem. It removes seasonal factors. Time-series regression on seasonally adjusted data can capture hidden patterns. Predicted values on seasonally adjusted data are then converted back to actual values. This process can significantly improve accuracy of predictions. It is noted that to make seasonally adjusted sales forecasting works, multi-year series data is required. At least three years data is recommended.
Sales Linear Growth Estimation by Linear Time-Series Regression
Linear growth can be best understood with linear time-series regression. Linear regression formula is stated as "Y = a + b * X" where "b" is the periodic average increase/decrease amount. "b" is the "Linear growth". The red line in the following figure shows the linear regression line. It represents the exact linear average growth amount. The gradient "b" of the red line is the linear growth. This method of growth computation is best for linear growth data. If data is not linear, this should be used with caution. "Linear growth %" is the percentage ratio of the "b" value over the average of input series data.
Rule-based Customer and Sales Trend Detection Methods
When the number of KPIs (or segments or products) is large, visualization methods are not feasible to apply. Rule-based automation can be used to detect trends automatically. RME-EP is a rule engine which supports various time-series regression and statistical functions. It's a perfect platform for automation of time-series trend detection. For example, the followings can be implemented with RME-EP;
This type of rules can be easily implemented with RME-EP rule engine. It requires KPI data to be stored as a relational database table. In addition, you need to learn SQL style coding. RME-EP is available from CMSR Data Miner / Machine Learning Studio. The following is an coding example. CMSR Download has the full code in the model "KPI Trends Detection Scan" of the "Demo" project. The model implements some above detections. RME-EP can update KPI database table and/or write into a CSV file. More example codes can be found here.
/* * If the difference between the current value and the predicted value is bigger than * rate-times of the regression average absolute errors. */ DECLARE "State-BigJump" AS STRING OUTPUT INITIAL VALUE 'No'; // evaluation status for big jump, if, 'Yes' RULE BigJumpDetection: IF ABS("Current" - TIMESERIES(1, (LINEAR, EXPONENTIAL), 1, "Series")) > TIMESERIES(AVG_ABS_ERROR, (LINEAR, EXPONENTIAL), 1, "Series") * 1.5 THEN SET "State-BigJump" AS 'Yes' END;
Excel Sales Trend Analysis and Sales Forecasting: Excel Group-by Add-in Tool
Rosella Group-by Excel Add-in provides powerful simple-to-use tools for trend analysis. It combines groupby aggregation with time-series predictive modeling. It employs powerful non-linear regression. Currently it supports 16 different mathematical functions using advanced function fitting algorithms. The following figure shows an example of Rosella Groupby Add-in reports with time-series analysis. (For full-size view, click the image.)
Columns "MTH1" to "MTH7" indicate recent actual trend values. Columns "Next 1" to "Next 3" show projections for the coming months. The "Chart" column displays them in charts. Notice that charts show that projections are not necessarily linear! The Addin automatically detects a most suitable function for you. Actual function description can be viewed from the notes. Prognostics columns provide information showing fitness and trends. The "Signal" column indicates which row items to pay attentions in three colors: green (for good), red (for bad), and yellow (for warning).
The following figure shows highlighted cells. It is based on column-wise high- and low-values. Red indicates high-values and blue shows low-value cells. Color intensity is based on relative cell-values. (For full-size view, click the image.)
Segment Time-series Trend Similarity / Correlation Analysis
The following figures show seven time-series trend charts. Consider the three time-series trends of the left figure. The first two series have identical positive growing trends. As the first series values rise, the second series values also rise. The third has opposite trend. As the values of the first series rises, the values of the third series decrease. This type of co-relationship is very common in business data. These three are correlated! The first and the second are positively correlated. The third is negatively correlated with the first and the second. Now consider the two series of the middle figure. Although not related in linear fashion, two values rise and fall together. That is, they are also closely correlated.
Consider the two series of the right figure. Notice that the second series values resemble the first series values with one period later. That is, one period time shift or time lag pattern. Second pattern occurs after one period lag. This this is also very important co-relationship.
Level of correlation can be measured in terms of correlation coefficients. It ranges between -1 and 1. If there is perfectly positive correlation, it is 1. If there is perfectly negative correlation, the value is -1. If no correlation, it is 0. The following chart shows a correlation matrix table of groups/segments. Red color indicates positive correlation. Blue is for negative correlation. Intensity of color is based on (absolute) correlation coefficients. With "Contrast" filters, most important correlations can be identified easily.
The following figure shows top 10 most correlated other groups of a selected group. It can be used to identify other groups with similar or opposite time-series trend.
Time-series Regression Function for Excel Formula with Seasonal Adjustment
Rosella Groupby Excel Add-in tool also provides the following time-series regression function that can be used in cell formula specification;
=RGTSREGRES(1, $D$1:$D$45, "01D",,4 ...)
This incorporates time-series regression with seasonal adjustment and smoothing (moving average and exponential) with the following parameters;
Examples Of Trend Forecasting with Seasonal Adjustment in Excel
The following codes can be used to produce long term forecasting work sheet with seasonal adjustment. The screen image shows example output of these formulas. It's very easy to do!
Download: Group-by Excel Addin Tool for Trend Analysis and Forecasting
For one year free license copy of this Excel Addin, click Download Rosella Excel Addin. This features the following tools (Last updated: May 11, 2018);
Download: Time-Series Regression Models for Android
For Time-Series Regression Models and Downloads, read Time-Series Regression Models for Android.