2. Final Formula
• 2nd Last step
User Score= (1*Facebook Score) +
(0.75*Google+ Score) + (0.5*Twitter Score) +
(0.25*LinkedIn Score)
• Last step
All user score will be brought down to 0-99 scale
by using percentile method. So the user with
highest score will have 99 iCandy score.
3. General Approach
• Reach has been calculated based on number of my
followers and average number of 2nd followers
• A log values is taken from the average number of 2nd
level followers.
• Inactive accounts and listeners are removed but
listeners are added as whole number to the final reach
value .
• Power of activity is calculated based on total activity
performed by self and the ratio of (total activity
performed by others favoring self)/(total activity
performed by others)
4. General Assumptions
• All analysis is done for Indian user only
• Types of activities have been restricted to one
given in the problem
• For each platform, reach has been restricted
to 2nd level of followers
• All calculations consider public posts and
activities.
• Certain weights to activities have been given
based on guess work
5. Reach Calculation
• Reach = (Connections)^(Log10(Average number of
2nd level connections)) + Listeners
• Connections = Total Connections – (Inactive +
Listeners)
• Total Connection: All accounts that follow me and
who get my activity news via pushed
notification/new feeds
• Inactive: All accounts with no activities in last 3
months
• Listeners: All account who follows me but never
perform any action
6. Platform-wise assumptions for Reach
• Facebook: All friends are followers and get
notification + Independent followers
• Twitter: All followers(Doesn’t allow
customization)
• LinkedIn: Connected associates + Independent
Followers
• Google+: All followers get notification
7. Facebook
Activity
• My activity: total number of posts +
(0.5) Automatic post on profile change +
(0.8) total no of wall post on other’s wall
• Followers activity 1: number of comments +
(0.2) number of likes
• Followers activity 2: (0.3) number of wall post on my wall
• Activity power = {My activity * (Average number of Followers activity 1 for
me/Average number of Followers activity 1)} + (Average number of
Followers activity 2 for me/Average number of Followers activity 2)
• Net activity power = (1*Activity Power of last 24 hours) + (0.5*Activity
Power of last week) + (0.25*Activity Power of last month) +
(0.125*Activity Power of ever before)
Influence
• Final Score = Reach*NetActivityPower
8. Google+
Activity
• My activity: total number of posts +
(0.5) Automatic post on profile change
• Followers activity 1: (1.2* number of shares) + number of
comments + (0.2*number of likes)
• Activity power = {My activity * (Average number of
Followers activity 1 for me/Average number of Followers
activity 1)}
• Net activity power = (1*Activity Power of last 24 hours) +
(0.5*Activity Power of last week) + (0.25*Activity Power of
last month) + (0.125*Activity Power of ever before)
Influence
• Final Score = Reach*NetActivityPower
9. Twitter
Activity
• My activity: (1*OriginalTweets) + (0.3*QuatedRT) + (0.2*AutoRT)
• Followers activity 1: number of RTs
• Followers activity 2: number of Mentions
• Activity power = {My activity * (Average number of Followers
activity 1 for me/Average number of Followers activity 1)} +
(Average number of Followers activity 2 for me/Average number of
Followers activity 2)
• Net activity power = (1*Activity Power of last 2 hours) +
(0.5*Activity Power of last 24 hours) + (0.25*Activity Power of last 2
days) + (0.125*Activity Power of ever before)
Influence
• Final Score = Reach*NetActivityPower
10. LinkedIn
Activity
• My activity: total number of posts +
(0.5) Automatic post on profile change
• Followers activity 1: number of comments + (0.5*number
of likes)
• Activity power = {My activity * (Average number of
Followers activity 1 for me/Average number of Followers
activity 1)}
• Net activity power = (1*Activity Power of last 2 days) +
(0.5*Activity Power of 2 week) + (0.25*Activity Power of
last month) + (0.125*Activity Power of ever before)
Influence
• Final Score = Reach*NetActivityPower
11. Normalizing each platform
• Each score for each platform is normalized
using
• This will produce each score on scale on 0 -1.
We then multiply by 100 and get all scores to
a 0 – 100 scale.
• These scores will be plugged into the final
score and then percentile method will rank
the users.