Skip to main content

Table 1 Features of human communication used in our analysis. Our feature types - Intensity (I), Active Periods (AP), Inter-event time (IET), Temporal Stability (TS), Bursty Cascades (BC), Distribution of bursty cascades (DBC), differences in daily patterns (DP) and clusters for weekly activity

From: Estimating tie strength in social networks using temporal communication data

Type

Variable

Name

Cluster

Description

I

w

Number of calls/contacts

\(C_{1}\)

Late night and early morning

I

l

Total call duration

\(C_{2}\)

Monday early morning

I

Average call duration

\(C_{3}\)

Monday early morning

I

r

Reciprocity

\(C_{4}\)

Weekday 7 am

AP

\(a_{d}\)

Active days

\(C_{5}\)

Weekday afternoon

AP

\(a_{h}\)

Active hours

\(C_{6}\)

Weekday evening

IET

τ̄

Mean IET

\(C_{7}\)

Weekday early morning

IET

\(\sigma _{\tau }\)

Std. Dev. of IET

\(C_{8}\)

Thursday early morning

IET

B

Burstiness Coefficient

\(C_{9}\)

Weekend evening

IET

\(\bar{\tau }_{R}\)

Average Relay Time

\(C_{10}\)

Weekend morning

TS

Relative freshness

\(C_{11}\)

Saturday Morning

TS

age

Age

\(C_{12}\)

Weekend afternoon

TS

TS

Temporal Stability

\(C_{13}\)

Saturday late afternoon

BC

\(N^{E}\)

Number of busty events

\(C_{14}\)

Sunday morning

BC

Ē

Average calls per bursty event

\(C_{15}\)

Sunday afternoon

BC

\(\sigma ^{E}\)

Std. Dev. of event distribution

\(C^{*}\)

Vector of clusters

BC

\(CV^{E}\)

CV of event distribution

  

DBC

Avg. interaction time

  

DBC

\(\sigma _{t}\)

Std. Dev. of interaction times

  

DBC

log(T)

Test statistic for

  

DP

JSD

Differences in daily behaviour