Scribd Logo
Upload

Welcome to Scribd!

  • 40 views

    Problem: To Increase The MER of DRTV From The Current Level On The Given Dataset

    Uploaded by

    Pranjal Jaiswal
    The document discusses increasing the media efficiency ratio (MER) from current levels using a dataset of 532 entries with 11 variables. Graphical inference and k-means clustering are used to understand relationships between MER and variables like day, month, time slot, and content. K-means clustering identifies clusters, with one cluster having a higher MER of around 1.25. Recommendations include terminating advertisements in time slots 4, 7, and 8, as removing those data points increased MER.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd

    Problem: To Increase The MER of DRTV From The Current Level On The Given Dataset

    Uploaded by

    Pranjal Jaiswal
    40 views5 pages
    The document discusses increasing the media efficiency ratio (MER) from current levels using a dataset of 532 entries with 11 variables. Graphical inference and k-means clustering are used to understand relationships between MER and variables like day, month, time slot, and content. K-means clustering identifies clusters, with one cluster having a higher MER of around 1.25. Recommendations include terminating advertisements in time slots 4, 7, and 8, as removing those data points increased MER.

    Original Description:

    analysis on case study

    Original Title

    Problem

    Copyright

    © © All Rights Reserved

    Available Formats

    DOCX, PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document discusses increasing the media efficiency ratio (MER) from current levels using a dataset of 532 entries with 11 variables. Graphical inference and k-means clustering are used to understand relationships between MER and variables like day, month, time slot, and content. K-means clustering identifies clusters, with one cluster having a higher MER of around 1.25. Recommendations include terminating advertisements in time slots 4, 7, and 8, as removing those data points increased MER.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    40 views5 pages

    Problem: To Increase The MER of DRTV From The Current Level On The Given Dataset

    Uploaded by

    Pranjal Jaiswal
    The document discusses increasing the media efficiency ratio (MER) from current levels using a dataset of 532 entries with 11 variables. Graphical inference and k-means clustering are used to understand relationships between MER and variables like day, month, time slot, and content. K-means clustering identifies clusters, with one cluster having a higher MER of around 1.25. Recommendations include terminating advertisements in time slots 4, 7, and 8, as removing those data points increased MER.

    Copyright:

    © All Rights Reserved
    Download as DOCX, PDF, TXT or read online from Scribd
    Download as docx, pdf, or txt
    You are on page 1of 5

    Problem: To increase the MER of DRTV from the current level

    on the given dataset.


    Outline of Issues: I am assuming that our data set is exhaustive and our
    inferences are only based on those 532 data points. This case study
    identifies a scenario which is supposed to give a higher MER (greater than 2)
    based on the findings from given datasets.
    Theory Used: We find the correlation between the 11 variables which are
    assumed to affect our MER, using graphical inference methods and K-means
    clustering.
    Understanding of Data: In the data set we have 532 entries with 11
    variables so if we classify data with respect to each and every variable the
    distinctiveness factor would be high so predicting a pattern would be
    improbable so we into cut down the factors . While interpreting data it could
    be seen that I have omitted few variables (e.g. time, order). The same is
    done because when I considered the Time,slot and day. The frequency
    wouldnt change when we exclude the Time.slot.
    Method: Firstly, I glance the data set and found that MER is the ratio of sales
    versus spends. As MER ranges from 0-6.46, I chunked them into 5 groups
    ( which are 0-0.5,0.5-1.0,1.5-2.0 and greater than 2).Now, I plot the
    histograms of MER v/s day, MER v/s month, MER v/s Time.Slots and MER v/s
    Content of the infomercials (CREATIVE).

    MER V/s Monday


    On X-axis, we have the chunks (0.5 denotes the range of MER from 0-0.5 and
    1 denotes the range of MER from 0.5-1.0 and so on) of MER and on Y-axis,
    frequency of these MER on Monday. Similarly the other plots are plotted for
    rest of days.

    MER V/s October


    On X-axis, we have the chunks (0.5 denotes the range of MER from 0-0.5 and
    1 denotes the range of MER from 0.5-1.0 and so on) of MER and on Y-axis,
    frequency of these MER on October. Similarly the other plots are plotted for
    rest of months.

    MER V/s Timeslot1


    On X-axis, we have the chunks (0.5 denotes the range of MER from 0-0.5 and
    1 denotes the range of MER from 0.5-1.0 and so on) of MER and on Y-axis,
    frequency of these MER on Time.slot1. Similarly the other plots are plotted
    for other 7 Time.slots.

    MER V/s CREATIVE


    On X-axis, we have the chunks (0.5 denotes the range of MER from 0-0.5 and
    1 denotes the range of MER from 0.5-1.0 and so on) of MER and on Y-axis,

    frequency of these MER on CREATIVE.A. Similarly the other plots are plotted
    for rest types of infomercials.
    In MER v/s Time.slots plot, I found that for Time.slot 4, 7 and 8 frequency of
    MER (greater than 2) is 0. I removed those Time.slots data points from the
    datasets and again plot the above graph to find some pattern in the data.
    Up till now we interpreting data on excel to look further, we used K-means
    clustering to recognize the cluster with MER greater than 2. To implement Kmeans clustering we need to find numbers of clusters for that we calculated
    the value of within groups sum of squares, to find value=4 to be optimum.
    Clusters we found after K-means clustering is as follows:
    Day
    3.5124
    3.4889
    3.9138
    2.9886

    Time.Slot
    4.0187
    3.6933
    2.1034
    3.000

    Area
    31.1062
    28.4578
    7.2069
    28.2977

    Channel
    12.5188
    41.9822
    76.8276
    72.8864

    MER VALUE
    0.8141
    0.9031
    0.8710
    0.6978

    For the second cluster MER value seems to highest but I further applied the
    K-means clustering within this clusters because the variance of this clusters
    could be large due to which the values we got wouldnt make any sense. On
    application of K-means clustering in second clusters we get.

    Day
    3.9
    3.5227
    3.556

    Time.Slot
    3.4
    4.1477
    3.2889

    Area
    35.1
    35.9091
    19.533

    Channel
    53.2500
    36.6705
    49.2334

    MER VALUE
    1.2515
    0.8496
    0.8529

    2.7436
    3.9394

    2.7949
    2.2727

    14.9231
    32.7273

    35.4615
    47.12

    0.9262
    0.8867

    In the above clustering methods we have assumed that although the


    elements are discreet .We can assign nearest integer value.
    Reason to choose to particular method:
    Our objective was to increase the MER for which we see how MER co-varies
    with other variables. For that I plotted histograms of MER v/s day, MER v/s

    month, MER v/s Time.Slots and MER v/s Content of the infomercials
    (CREATIVE). These helped us to infer the factors that are inversely
    proportional to MER.
    Recommendation:
    1. Advt. that is being broadcasted in the Time.slots of 4, 7 and 8 should
    be terminated. As after removing these points the MER increased
    significantly.
    2.

    You might also like