Supply Chain Challenges And Forecasting Techniques

Vendor management or Vendor list selection

Three possible challenges that can be identified from the given supply chain diagram are Vendor management or Vendor list selection, Optimization of Transportation costs and Forecasting of Inventory Stocks.

Selection of the appropriate or best vendors can have far reaching consequences on the supply chain. It can influence the cost of operations, risk and quality of the products. The task of choosing the right vendor from the available options in the market is therefore crucial. This requires keen market insight. The objective is mainly to select those who would offer transparency in business, will be easy to communicate with when it comes to planning strategy and most importantly offer quality products (Bozarth  and Handfield  2019) .

Yet another issue is to identify the best options whereby transportation costs could be reduced in terms of time and capital. The challenge in this case is to identify the best routes in times of cost as well as time efficiency. Transportation of goods forms a key aspect of supply chain management which in turn may affect product quality and delivery as well as the over all costs (Christopher  2016).

Inventory management deals with the challenge to plan and procure goods for business. An efficient supply chain takes into account the demand of goods in the market and plans the supply to meet the demand. It is crucial to be able to tell how much the demand for a product will be on the market and hence how much stock one must keep in the inventory to ensure that the inventory is neither surplus nor falls short of the supply (Christopher, 2016).

Consumer Market Surveys are tools to gain market insights by companies from customers themselves. The survey involves asking people about their purchases and preferences. The surveys can be conducted via telephones or through personal interviews or even by questionnaires that the consumer is requested to fill out and submit either via mail or online. The responses are then analyzed using statistical tools and methods to gather market insight and validate conjectures regarding the market. 

The Naïve forecasting method is the simplest forecasting technique of all. The method operates under the assumption that what happens today is expected to happen tomorrow as well. This means it assumes that the observations of the current time point as the forecast for the subsequent time period. It disregards any causality or influence of observations from time points preceding the current time point. So if observed Y at time point t is Y_t then as per this method the forecast for time point t+1 is Y_t (Montgomery, Jennings  and Kulahci  2015). 

Optimization of Transportation costs

The simple linear regression technique identifies two variables, one dependent and the other independent based on whom a model establishing the linear relationship between the two is identified. Simple linear regression is denoted by a straight line where the Y axis denotes the dependent and X denotes the independent variable. The regression line is the line of best fit which is line for which sum of squared errors from each data point (defined by a pair of observed (x, y)) is least.  Let Y= a+ bX,  be the regression line. Then this equation gives a value of Y for each value of X. This is how the forecasting works (Chatterjee and Hadi 2015).

Period	Actual Value	Naïve Forecast	Error	Absolute Error	Percent Error	Squared Error
January	10	N/A	N/A	N/A	N/A	N/A
February	12	10	2	2	16.66667	4
March	16	12	4	4	25	16
April	13	16	-3	3	23.07692	9
May	17	13	4	4	23.52941	16
June	19	17	2	2	10.52632	4
July	15	19	-4	4	26.66667	16
August	20	15	5	5	25	25
September	22	20	2	2	9.090909	4
October	19	22	-3	3	15.78947	9
November	21	19	2	2	9.52381	4
December	19	21	-2	2	10.52632	4
			0.818182	3	17.76332	10.09091
			BIAS	MAD	MAPE	MSE
		Standard error  (Square Root of MSE) =3.176619

Data on sales of cosmetics for the year 2017 for a cosmetic company was used to forecast units sold per month using the technique showed in the given diagram. The computation was done by finding the sum of product of observed value with the weights 0.2, 0.3 and 0.5 respectively for the preceding three years of the current forecast.

The following table shows the data and the corresponding forecasts.

Month	Actual Units sold (in 000)	Three Period Moving Average	weights
January	155		0.2
February	168		0.3
March	159		0.5
April	162	161
May	178	162
June	177	169
July	189	174
August	198	183
September	205	191
October	215	200
November	220	209
December	210	216
January	214	214	(forecasted)
February	214	214	(forecasted)
March	213	213	(forecasted)
April	214	214	(forecasted)
May	214	214	(forecasted)
June	214	214	(forecasted)

The forecast for the next January is 214 which is an increase from that observed in January of current year. Computing further forecasts using forecasted values, values are seen to become stable. 

The same data as the previous part was used to forecast using Moving average method in this part. The moving averages were computed by taking weighted averaging using given weights over the past three years of the time point for which forecast is being made. The data therefore does not have forecasts for the first three time points.

Month	Actual Units sold (in 000)	Weights	Three Month Weighted Moving average
January	155	0.76
February	168	0.18		Calculation of Weighted Moving Average (Period=3)
March	159	0.06		Sum/ Sum of weights(=1)
April	162		157.58	117.8	30.24	9.54
May	178		166.02	127.68	28.62	9.72
June	177		160.68	120.84	29.16	10.68
July	189		165.78	123.12	32.04	10.62
August	198		178.48	135.28	31.86	11.34
September	205		180.42	134.52	34.02	11.88
October	215		191.58	143.64	35.64	12.3
November	220		200.28	150.48	36.9	12.6
December	210		207.7	155.8	38.7	13.2
Next Period		215.6	163.4	39.6	12.6
Sum of Weights	1

The forecast for the January of next year is 215.6. So there has been  positive change or increase in units sold when compared to observation of January of current year.

The completed regression analysis for the given diagrams and tables is given in the following table. The estimate of slope and intercept was computed using the formula as provided in the example:

	Actual Value (Y)	Period number X	XY	X^2
	74	1995	147630	3980025	Slope=	b=	10.53571429
	79	1996	157684	3984016	Intercept=	a=	-20951.5
	80	1997	159760	3988009
	90	1998	179820	3992004
	105	1999	209895	3996001
	142	2000	284000	4000000
	122	2001	244122	4004001
Total	692	13986	1382911	27944056
	X	Predicted Y
	2002	141

The trend line was thus found to be given by the equation:

y= 10.54x- 20951.5

(b) The following analysis was based on collected data, on distance travelled by a vehicular unit and the fuel consumed by the same. The fuel consumed is the dependent variable and the distance travelled is the independent variable. The following table shows the computations.

	Distance Travelled (X)	Fuel Consumed (Y)	XY	X^2	Y^2
	75000	12.5	937500	5625000000	156.25
	75300	9.5	717142.9	5670090000	90.7029478
	75600	12.0	907200	5715360000	144
	75900	11.1	843333.3	5760810000	123.45679
	76200	9.4	714375	5806440000	87.890625
	76550	11.3	864274.2	5859902500	127.471384
	76820	9.0	691380	5901312400	81
	77110	10.0	771100	5945952100	100
	77460	13.0	1004111	6000051600	168.038409
	77690	8.2	638167.9	6035736100	67.4744898
SUM	763630	105.9774919	8088584	5.8321E+10	1146.28465
r		-0.315
Slope=	b=	-0.00055
Intercept=	a=	52.6730

The regression equation was found as y=52.6730 – 0.00055x. So increase in distance travelled would correspond to a decrease in fuel consumed by 0.00055 units.

Reference

Bozarth, C.C. and Handfield, R.B., 2019.operations and supply chain management.

Chatterjee, S. and Hadi, A.S., 2015. Regression analysis by example. John Wiley & Sons.

Christopher, M., 2016. Logistics & supply chain management. Pearson UK.

Montgomery, D.C., Jennings, C.L. and Kulahci, M., 2015. Introduction to time series analysis and forecasting. John Wiley & Sons.