Upload
rajib-layek
View
223
Download
0
Embed Size (px)
DESCRIPTION
Citation preview
TEAM: SOA
RAJIB LAYEK & ARINDAM ROUTH
– IIT KHARAGPUR
1Problem Statement 1: Which two products the Company should launch first, and why?
Computech Technologies (P) ltd should launch CPA & CFM or CPBA & CFM as of now to maximize their profit and minimize risk.
Reason: Irrespective of any program they are incurring fixed cost of 2.1048 Cr INR. So, the decision to launch which two products directly depends on variable cost of those two programs.
Now if we assume that 3 month course can be done repeatedly 4 times in a year in the existing infrastructure then for CBA faculty cost will be 440,000 INR for 3 month (110 hr. * 1000 INR per hr.*4 times in a year. So, revenue from CBA per year per person is 480,000 INR (120,000 * 4 Times a year).
For CBA+CFM, TC: 21,048,000 (FC) + 440,000 (VC for CBA) + 396,000 (VC for CFM) = 21,884,000. Revenue from these two product will be 480,000 (CBA for 1 yr.) + 480,000 (CFM for 1 yr.) = 960,000.
Break-Even Analysis: So, min no of students required for Break-Even 21,884,000 /960,000 = 22.796 ≈ 23.
In similar way min no of students required for all the combination can be calculated. And the least no of students are coming for CBA+CAPB or CBA+CFM. For detailed calculation see Appendix A.
Risk Aspect: for CBA+CFM combo 23 students required for whole year which means 23/ (4*2) ≈ 3 students per course which is a safe assumption.
Qualitative Analysis: As the BRAND is not established yet people won’t like to invest 6.2 Lakh for 2 year course at first. So, it’s better to start with 3 month course and then after reasonable brand awareness they can launch other long term product.
2
APPENDIX 1: BREAK-EVEN ANALYSIS IN EXCEL Table 1:
ABC College
Product Details
Product Name
Course Duratio
n
Functional Area
Product Unit Price
Delivery Hrs
Faculty Cost per Hr.
Total Faculty Cost per course
Total Faculty
Cost in one year
Delivery hour
Per M
Faculty Cost Per
Month
PGDBA 12 Technology 360,000₹ 460 1,000₹ 460,000₹ 460,000₹ 38 38,333₹
PGDMA 24 Management 620,000₹ 800 900₹ 720,000₹ 360,000₹ 33 30,000₹
CBA 3 Technology 120,000₹ 110 1,000₹ 110,000₹ 440,000₹ 37 36,667₹
CPBA 3 Technology 120,000₹ 110 1,000₹ 110,000₹ 440,000₹ 37 36,667₹
CFM 3 Management 120,000₹ 110 900₹ 99,000₹ 396,000₹ 37 33,000₹
PGCBA 12 Management 360,000₹ 460 900₹ 414,000₹ 414,000₹ 38 34,500₹
Table 2:
Name CodeCourse Duration Cost Revenue
PGDBA Course 1: C1 12 460000 360000PGDMA Course 2: C2 24 360000 310000CBA Course 3: C3 3 440000 480000CPBA Course 4: C4 3 440000 480000CFM Course 5: C5 3 396000 480000PGCBA Course 6: C6 12 414000 360000
FC 21,048,000
Table 3:
No of Combination
Course Combination
Variable Cost Total Cost Total Price
No. of Students in a year
No. of Students per course
Profit for min 50 students per year
1 C1+C2 820000 21,868,000 670000
32.64 11,632,000
2 C1+C3 900000 21,948,000 840000
26 20,052,000
3 C1+C4 900000 21,948,000 840000
26 20,052,000
4 C1+C5 856000 21,904,000 840000
26 20,096,000
5 C1+C6 874000 21,922,000 720000
30 14,078,000
6 C2+C3 800000 21,848,000 790000
28 17,652,000
7 C2+C4 800000 21,848,000 790000
28 17,652,000
8 C2+C5 756000 790000 17,696,000
321,804,000 28
9 C2+C6 774000 21,822,000 670000
33 11,678,000
10 C3+C4 880000 21,928,000 960000
22.842
2.85
26,072,000
11 C3+C5 836000 21,884,000 960000
22.796
2.849
26,116,000
12 C3+C6 854000 21,902,000 840000
26 20,098,000
13 C4+C5 836000 21,884,000 960000
22.796 2.849 26,116,000
14 C4+C6 854000 21,902,000 840000
26 20,098,000
15 C5+C6 810000 21,858,000 840000
26 20,142,000
For calculation and formula see the attached excel file.
Problem statement 2: Based on analysis of the research data provided by the Marketing research company, work out the profile of the kind of Consumer, the Company must focus on.
Objective is to build a predictive logistic regression model which will help client to identify the customer profile. For this as we have already seen that we are concerned about customer for only 3 month course (CBA/CPBA.CFM) we have taken a sample of 341 data points which include all the data points who enrolled for these three program also 74 randomly chosen data point who didn’t apply for any of the program.
On that sample points binary logistic regression is run and output is interpreted in the below section.
4
Here CBA is dependent variable with value 0 or 1 which represent customer will not Enroll (0) or Enroll (1) for CBA/CPBA/CFM. All the other characteristics of the customer is taken as independent variable. Among these apart from Salary, Work-Ex and Age all are categorical variable.
Forward LR method is taken as no prior research work has been done in this regard.
INTERPRETATION OF OUTPUTTable1: 341 sample points are taken. From the whole set of data for preparing logistic regression for CBA/CPBA/CFM enrollment.
Case Processing Summary
Unweighted Casesa N Percent
Selected Cases Included in Analysis 341 100.0
Missing Cases 0 .0
Total 341 100.0
Unselected Cases 0 .0
Total 341 100.0
a. If weight is in effect, see classification table for the total number of
cases.
Table 2: Dependent variable CBA: Yes & No. (Whether people will enroll for CBA/CPBA/CFM or not.)
5Dependent Variable Encoding
Original Value Internal Value
N 0
Y 1
Table 3: Different label of specification, Functional area, level, Highest Qualification are coded.
6
Table 4: CBA/CPBA/CFM enrolled 266 person and not enrolled 75 person.
Classification Tablea,b
Observed
Predicted
CBA
Percentage CorrectN Y
Step 0 CBA N 0 75 .0
Y 0 266 100.0
Overall Percentage 78.0
a. Constant is included in the model.
b. The cut value is .500
Table 5: As we have done forward stepwise method for logistic regression, Initial
model includes only constant term (excluding all other predictors) and the log-
likelihood of this base model is 359.301.
7
Iteration Historya,b,c
Iteration -2 Log likelihood
Coefficients
Constant
Step 0 1 360.578 1.120
2 359.302 1.261
3 359.301 1.266
4 359.301 1.266
a. Constant is included in the model.
b. Initial -2 Log Likelihood: 359.301
c. Estimation terminated at iteration number 4 because parameter estimates changed by less than .001.
Table 6: 78% time the model predicts correctly that the person will enroll for CBA/CPBA/CFM.Classification Tablea,b
Observed
Predicted
CBA
Percentage CorrectN Y
Step 0 CBA N 0 75 .0
Y 0 266 100.0
Overall Percentage 78.0
a. Constant is included in the model.
b. The cut value is .500
Table 7: Value of the constant (b0) = 1.266
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 0 Constant 1.266 .131 93.769 1 .000 3.547
Table 8: Residual Chi-square is calculated with <.001 level of significance which
indicates co-efficient of other variable are significant which means addition of one
or more variables to the model will significantly affect its predictive power.
8
9Table 8: -2Log-Likelihood (-2LL) has decreased significantly to 106.994 from the initial -
2LL value of 359.301. Lower value of -2LL is indicating that model is predicting more
accurately as 2LL indicates unexplained data.
Model Summary
Step -2 Log likelihood Cox & Snell R Square Nagelkerke R Square
1 248.067a .278 .427
2 171.609b .423 .650
3 129.283b .491 .753
4 113.286b .514 .789
5 106.994b .523 .803
a. Estimation terminated at iteration number 5 because parameter estimates changed by less than .001.
b. Estimation terminated at iteration number 20 because maximum iterations has been reached. Final solution cannot be
found.
Table 9: At step 1 Age variable is included in the model. Wald statistic for Age is 65.499 with significance label less than .001. Wald statistic follows a Chi square distribution and tells us b-coefficient of Age is significant. And co-efficient for Age (b1) = -0.113
Exp(B)= 0.893 which is less than 1 which indicates if Age increases probability for Enrollment decreases. (If the value of Exp(B) would be greater than 1 then it would increase with increase in Age )
In Step 2, 3 and 4 Educational Qualification, Functional Area and Work-Ex is included but result is not significant. So only step1 in considered. So, no other predictor is taken for the model.
10
Table 10: Change in -2LL for Age is 111.234 with significance label of <.001 which means it has a significant effect on the predictive ability of the model.
Model if Term Removed
Variable Model Log Likelihood
Change in -2 Log
Likelihood df Sig. of the Change
Step 1 Age -179.650 111.234 1 .000
Step 2 Age -118.584 65.558 1 .000
Edu -124.033 76.458 17 .000
Step 3 Age -83.123 36.963 1 .000
Func_Area -85.805 42.326 17 .001
Edu -93.478 57.674 17 .000
Step 4 Age -82.133 50.980 1 .000
Func_Area -84.296 55.306 17 .000
Work_Ex -64.641 15.997 1 .000
Edu -90.499 67.712 17 .000
Step 5 Age -79.534 52.073 1 .000
Func_Area -82.120 57.245 17 .000
Work_Ex -63.837 20.680 1 .000
Salary -56.643 6.291 1 .012
Edu -90.055 73.117 17 .000
Table 11: Histogram of the predicted probabilities of the people enrolling in the
CBA/CPBA/CFM program.
11
This graph tells us how well the predictive model is. As max clusters are at each end the model and no of N in right hand side and no of Y in left hand side is very less the model is reliable.
So, apart from age no other characteristics can be predicted as a significant predictor for the logistic regression which will help client to identify customer profile.
Co-efficient of constant and Age (B0) and (B1) for the model is 5.622 and -0.133 respectively.