HW 1
HW 1
Before answering make sure you modify the data based on f factor. Calculate f factor as f= max
[0.2, (Average of the last two digits of your ids)/100] , If you are a group
f = max(0.2,((82+71/2)/100))
f = max(0.2,(76.5/100))
f = max(0.2,0.765)
f = 0.765
You MUST use f in adjusting question data as directed in the questions BEFORE answering them
20
0
-200.5 1 1.5 2 2.5 3 3.5 4 4.5
-40
-60
Years
Adjusted Board-
ings(millions)
300
ings(millions)
200
Predicted Ad-
100 justed
0 Boardings(million
0 1 2 3 4 5 s)
Years
The fit plot (Years Line Fit Plot) displays how actual adjusted boardings compare with
predicted adjusted boardings throughout the years. The actual observed values appear
as blue dots on the graph while orange squares show the predicted values from the
model. A well-fitted model should demonstrate a strong correlation between predicted
values and actual values. The graph demonstrates a reasonable correlation between
actual and predicted values yet displays some deviation. The model operates well
when deviations are minor and randomly distributed but systematic deviations suggest
areas requiring improvement.
The Years Residual Plot visualizes the discrepancies between observed values and
their corresponding predicted values. The model's errors should be distributed
randomly around zero as shown by a lack of any distinct pattern in the residuals. The
residuals demonstrate a pattern because some values deviate substantially from zero.
The model appears to miss some data variations since its predictions do not account
for all patterns in the data. A visible pattern in the residuals suggests that the
predictive accuracy might be improved by using a different model or adding more
explanatory variables.
Before answering, Add 20xf to the years 89-90, 90-91, 91-92, and 2009-2010.
Q2. (40 points) Following is data for the last three years about the tonnage transported out
of a city by rail (given as excel file Hw1_seasonal_demand.xls ).
a. Plot the data and decide if we can assume data to be stationary and seasonal, why?
The patterns of the data shows peaks repeating pattern that’s we can assume it is
Seasonal vs Stationary
800.0
600.0
Demand
400.0
200.0
0.0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
month
seasonal but as demand is different thought out the time, that’s why this data isn’t
stationary.
b. Find the seasonality factor for each month (be careful about outliers).
c. Forecast for each month in next year.
d. Calculate MSE, MAD, and MAPE for your forecast.
Before answering Add 50xf to months 6, 7, 8, 17, and 18 to the data given in the excel
file.
Q3. (20 points) Assume that we have a city A and city B without a train link now. The
distance between the cities is 250 km and the population of the cities are 2.5+(0.5f) million
and 4+(-0.5f) million respectively. Based on past annual passenger demand for rail between
different cities with train link, we estimate the factor k as interval and this range is ( 0.0011,
0.0014). Using the gravity model estimate the annual rail passenger demand between the
cities, best and worst case. Would you establish the train link between the cites, why? (Make
sure you modify the data as given in the question)
Annual rail passenger demand for rail = Demand factor*(Population of City A* Population of
City B)/( Distance between the cities)^2
Population of City A = 2.5+(0.5*0.765) = 2.8825 million
Population of City B = 4-(0.5*0.765) = 3.6175 million
To be submitted:
- Your answers as MSWORD file.
- Imbed ALL the results and plots from excell into the word file- MSWORD file will be
graded
- Excel file used
- One of the group members can submit
- Late submission up to one day – 30% off the grade HW deserves. No submission after
one day