You signed in with another tab or window.Reload to refresh your session.You signed out in another tab or window.Reload to refresh your session.You switched accounts on another tab or window.Reload to refresh your session.Dismiss alert
A library to find the Probability Estimation of Navigation Paths and their Pattern Prediction.
The library helps in identifying the high probability trail path in a data. This navigation probability provides the means to analyze and predict the next link choice of unseen navigation sessions. Currently, the library allows three types of probability estimation from the path data -
State Probability
Transition Probability
Path or Trail Probability
State Probability -
The initial probability of a state is estimated as the proportion of times the corresponding state was requested by the user. This probability is obtained by dividing the number of times a state was browsed by the total number of states browsed.
Transition Probability -
The probability of a transition between two states is estimated by the ratio of the number of times the sequence was visited tothe number of total paths where the from page was visited.
Path or Trail Probability -
The probability of a trail is estimated by the product of the initial probability of the first state in the trail and the transition probabilities of the next transitions taken in a path. The chain rule is applied in order to compute all path probabilities.
# To find the state probabilitystate_probability = nv.state_probability(df, 'path')print(state_probability)
State State_probability0 D 0.1428571 A 0.3571432 C 0.2142863 B 0.285714
# To add the start and conversion values to the path (optional)df = nv.add_start_end(df,'path','conversions')print(df)
other_data path conversions0 1 [start, A, B, C, A, C, exit] 01 4 [start, B, D, B, A, exit] 02 5 [start, A, C, B, A, D, conversion] 1
# To find the transition probabilitytransition_df = nv.transition_probability(df, 'path')print(transition_df)
from_sitesection to_sitesection transition_probability B exit 0.000000 B conversion 0.000000 B D 0.333333 B A 0.666667 B C 0.333333 D exit 0.000000 D conversion 0.500000 D B 0.500000 D A 0.000000 D C 0.000000 start exit 0.000000 start conversion 0.000000 start B 0.333333 start D 0.000000 start A 0.666667 start C 0.000000 A exit 0.333333 A conversion 0.000000 A B 0.333333 A D 0.333333 A C 0.666667 C exit 0.500000 C conversion 0.000000 C B 0.500000 C D 0.000000 C A 0.500000
# To find the path probabilitypath_df = nv.path_probability(df, 'path', transition_df)print(path_df)
other_data path conversions path_probability0 1 [start, A, B, C, A, C, exit] 0 0.0123461 4 [start, B, D, B, A, exit] 0 0.0123462 5 [start, A, C, B, A, D, conversion] 1 0.024691
Additional functions -
# To convert the path column to string data typedf = nv.convert_to_str(df, "path")# To convert the path column to list data typedf = nv.convert_to_list(df, "path")
About
Library to find the Probability Estimation of Navigation Paths and their Pattern Prediction.
