Industrial Engineering : Forecasting

We can define forecasting based on time duration as:

• Short term forecasting for 1 to 3 month
• Intermediate-term forecasting for 3 to 12 months
• Long term forecasting for more than 1 year

Types of Demand Variation:

Trend Variation (T): It shows a long-term upward or downward moment in the demand pattern of the particular product.

Seasonal Variation: It shows a short-term regular variation related to a particular time of a day or day of a week.

Cyclic variation (C): It shows a wave-like demand variation for a larger period that is one year or more.

 Level or Constant (L): The demand remains almost constant with respect to some parameter.

Irregular variation (I): These variations are caused due to unusual circumstances which are not reflective of normal behavior. These may be due to govt policy changes, price hikes, a strike shut down, etc. These are always neglected while forecasting.

Classification of Forecasting Method

Time Series: In this method, past data is arranged in chronological order as a dependent variable and time as the independent variable. Based upon this past data, we need to project the demand in the future.

1. Past Average: In this method, the forecast is given by average or mean of the actual demand data of the previous period.
2. Simple moving Avg. or Simple rolling avg. (SMA): This method uses the past data and calculates the rolling avg. for a constant period. The fresh avg. is computed at the end of each period by adding the actual demand data for the most recent period and deleting the data for the older period. In this method as data changes from period to period, it is termed as moving avg. method.
3. Weighed moving avg. (WMA): WMA method uses past data similar to the simple moving average. The difference being it gives unequal weight to each demand data in such a manner that summation of all weight always equal to one. The most recent data is given the highest weight and the weight assigned to the oldest data will be the least.
4. Exponential smoothing: This method required only the current demand and forecasted value for the current period to give the next forecast. This method is the modified form of weighted moving avg. which gives weight to all the previous data and the weight assigned are in exponentially decreasing order.

Responsiveness:

• It indicates that the forecast has a fluctuating or swinging pattern.
• It is preferred for new products and for that no. of the period is kept small.

Stability:

• It means the forecast pattern is flat, smooth or has less fluctuation.
• It is preferred for old existing product and for that no. of the period is kept large.

Trend Projection in Forecasting: In the trend projection method of forecasting, the value of time series exhibits a long-term linear trend.

 Where, y₁ = The trend value,
            a = Intercept of the trend line
            b = Slope of the trend line.
            x = Independent variable

Alternatively, 

Casual Forecasting Method: In the casual method, we try to establish a cause and effect relationship between changes in the series level for the product and set up relevant explanatory variables. It can be divided into simple regression analysis and multiple regression analysis

Forecast Error: A forecast error is a difference between the actual or real and the predicted or forecast value of time series or any other phenomenon of interest. There are two methods which are given below.

• Mean Absolute Deviation (MAD): In this method, we calculate as the average of the absolute value of the difference between actual and forecasted values. The negative sign in this difference is ignored as overestimating as well as underestimate are both off-target and thus desirable, Where D_t = Actual value of demand for period
            F_t = Forecasted demand for period t
            n = Number of periods considered for calculating error
• Mean absolute percent deviation (MAPE): It is the average value of percentage error compared to actual demand. It is used to put an error in perspective because there is a difference between 40 out of 100 and 40 out of 1000.
• Mean Sum of Square Error (MSE): Average of all squares of all errors in the forecast is termed as the mean sum of the square method. It is written as
• Bias: It is a measure of overestimation or underestimation. If the bias value is positive, then it is an underestimation and if bias is negative, then it is an overestimation.
• Tracking Signal: It is the ratio of bias till n periods to mean absolute deviation till n periods. It is used to identify those items which do not keep pace with either positive or negative bias or trend.

You can avail of BYJU’S Exam Prep Online classroom program for all AE & JE Exams:

BYJU’S Exam Prep Online Classroom Program for AE & JE Exams (12+ Structured LIVE Courses)

You can avail of BYJU’S Exam Prep Test series specially designed for all AE & JE Exams:

BYJU’S Exam Prep Test Series AE & JE Get Unlimited Access to all (160+ Mock Tests)

Thanks

Team BYJU’S Exam Prep

Download  BYJU’S Exam Prep APP, for the best Exam Preparation, Free Mock tests, Live Classes.