Sie sind auf Seite 1von 23

Analysis Of Nutrition Facts In Food Products

Submitted by:
Submitted to: Keshav Kumar
Mr. Kaushal Kumar
Ankit Sehgal
Vinod Sahu
Tarun Yadav
Description

 Quick snack and fast food are part of our daily lives. We all are so relied on
them that we are always consuming them in the smallest gaps of our daily
activities. What is our interest is the “Nutrition Value” associated with them.
Nutrients we get from food consists of “Macronutrients” and
“Micronutrients”.
Almost every food having Agmark in India has the nutrition values written on
the package of the item. Along with the Major Nutrients i.e. Carbohydrates,
Fat and Protein ; we also have the information about Energy that the
consumable provide.
Objective

 Our Objective with this project is to establish a linear relationship of the


nutrients present in the packaged food item with the Energy level provided by
the product.
Data

The description of our data is as follows:


 Name: Name on the label of food product
 Energy: Energy in kcal per 100g of food item
 Protein: Proteins in g per 100g of food item
 Carbs: Carbohydrates in g per 100g of food item
 Fat: Fat in g per 100g of food item
 Sugar: Sugar in g per 100g of food item
 Others: Other constituents in g per 100g of food item
Link: https://drive.google.com/file/d/0B8mb-
qMcwb06dVhDNjVvOEtmS3c/view?usp=sharing
MODELLING

Dependent Variable Independent Variable


Energy Protein
Carbs
Fat
Sugar
Others
Coefficients
Anova
Model

 Energy = B0+ B1*Protein+ B2*Carbs+ B3*Fat+ B4*Sugar+ B5*Others + Residue


Where B0 = 10.99625
B1 = 2.34196
B2 = 3.83395
B3 = 9.15754
B4 = -0.18520
B5 = 0
Moreover the p-value in the above table shows that the independent variables
“Sugar - 0.5340” and “Others - ” are highly insignificant in our model.
Comparison

Energy vs Predicted Energy


1000

900

800

700

600

500

400

300

200

100

0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91

Energy Predicted Energy


Multicollinearity
Observation

 All the Variance inflation corresponding variables is < 2.5 except that of
“Others” variable.
 All the tolerance value of corresponding variables is > 0.4 except that of
“Others” variable.
 Moreover, The linear relation among “Others” and rest of the explanatory
variables can be clearly seen.
𝜆𝑚𝑎𝑥
 𝜅= > 1000
𝜆𝑚𝑖𝑛

 Outcome: Perfect Multicollinearity is present.


Autocorrelation
Heteroscedasticity
Observation

Result: The sig. value i.e. .029 < 0.05


So we reject out null hypothesis and accept the Alternative
hypothesis.
 Heteroscedasticity is present in our data
IMPROVEMENT

 Removed the redundant variable “Others” from the data


 Best model selection : Stepwise Selection

Dependent variables Independent Variables


Energy Protein
Carbs
Fat
Sugar
Final Outcomes
Comparison

Regression vs Original
1000

900

800

700

600

500

400

300

200

100

0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91

Energy Predicted Energy


FINAL INTERPRETATION

 Energy = B0+ B1*Protein+ B2*Carbs+ B3*Fat+ Residue


 Where B0 = 9.71271
B1 = 2.44540
B2 = 3.79049
B3 = 9.16854
The final model shows that :
 There is 2.44540 increase in Energy w.r.t per unit change in Protein content in
Food Item
 There is 3.79049 increase in Energy w.r.t per unit change in Carbohydrate content
in Food Item
 There is 9.16854increase in Energy w.r.t per unit change in Fat content in Food
Item
Trendline

Predicted Energy
1000

900

800

700

600

500

400

300

200

100

0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91
Softwares Used

 SAS University Edition


 IBM SPSS 22
 Microsoft Excel 2013
 R Studio(For verification only)
Thanks

 Kunal Dang
 Ashu Kush
 Priyanka Tewatiya
 Saksham Jain
 Sadashiv Gupta
 All our classmates for all the support

Thank You.

Das könnte Ihnen auch gefallen