Connector Pda Case Analysis
Essay Preview: Connector Pda Case Analysis
Report this essay

First, the group reviewed the dendrogram for significant â€śjumpsâ€ť from one cluster to another upwards along the lines. To familiarize our group with the data available, we ran a 9-cluster segmentation to visualize the cluster types by distance on dendrogram and discern cluster size. Running a 9-cluster segmentation delivers a reference dendrogram and cluster-size statistics. The initial 9-cluster dendrogram is illustrated in Exhibit 2. Additionally, the group ran a 12-cluster Exhibit 3, 6-cluster Exhibit 4, and 4-cluster Exhibit 5 segmentation. The exhibits are valuable in illustrating where disparate clusters join together as the number of total clusters decreases. Though the dendrograms can give an indication of appropriate grouping levels, the dendrogram is only a â€śfamiliarizationâ€ť step. This is where the second step goes further. Secondly, our group visualized the resulting data in a graph, resulting in Exhibit 6. On the x-axis, the total number of clusters was plotted, 1 through 9. On the y-axis, the distances from the dendrogram were plotted, 0 to 3. From this graph, the group searched for a distinctive change in the line, where the increase or decrease in cluster numbers has an identifiable effect upon the distance. This identifiable jump occurs at choosing 3 clusters. Additionally, progressing to 5 clusters provides negligible value-add to the segmenting process. Therefore, the optimal choice should be 4 clusters.Also, by running the k-means clustering analysis in R, graph (See Exhibit 7)Â illustrates the â€śbetween sum of squares/ total sum of squaresâ€ť ratio, which is basically a measure of the goodness of the classification that k-means has found. Usually, the higher the ratio, the better the model would be. However, the optimal choice of k will strike a balance between maximum compression of the data using a single cluster, and maximum accuracy by assigning each data point to its own cluster. In other words, Â to make our clustering that has the properties of internal cohesion and external separation, we decided the number of clusters by using the Elbow Method, which means choosing a number of clusters so that adding another cluster doesnt give much better modeling of the data. As shown in Exhibit 7, the “elbow” is indicated by the yellow dashed line. The number of clusters chosen should therefore be 4.Segmentation variable analysisBy mathematically doing clustering analysis, we arrived at a final 4-cluster (segment) result. At this moment, we still need to further analyze the segment variables to check whether our segmentation is practical and actionable in solving our business problems. We assigned our four clusters to â€śGroup Aâ€ť,â€ťGroup Bâ€ť, â€śGroup Câ€ť, â€śGroup Dâ€ť and tried to look at their scores across different variables. Using Nielsenâ€™s PRIZM consumer segments, we have named those four groups: Urban Achievers, Greenbelt Sports, Bedrock America,Upward Bounders(See Reference).Urban AchieverÂ Greenbelt SportsBedrock AmericaUpward BoundersPIM Master; Response required and requested on time-sensitive information; the multimedia segment; less-willing to pay than averageRequire remote access; Use messaging services; monitor emphasis;Most willing to pay; time-sensitive information segment (emergency); The Innovator; Communicates primarily by email; web access required;Based on our analysis on segmentation variables for 4 clusters. We found that those 4 groups have distinguishable preferences over PDA features, which proves that our segmentation is reasonable so far. InÂ Exhibit 8, it shows that Urban Achiever and Upward Bounders have both above average Innovator quality,which means that they are both very likely to adopt new technologies. On one hand, they share some similarities in cell phone usage, email and web access, and use their cell phones more often than other two Groups. On the other hand, Group A and D have dissimilarities in many other aspects, like price that they are willing to pay, ergonomic, PIM usage, remote access, etc. Therefore, at this point, Group A and D become our optimal choice since we can cater to two target markets which encompass similar needs and slightly dissimilar features. Specifically, by targeting these two groups, we can use basically similar marketing strategies with slight difference in price and functionality. Besides, Group A and D both have large market size, which is very important because segments should be large enough to be profitable. We didnâ€™t choose Group B and C not only because they have lower Innovator scores, but also majorly because of their small market size (See Exhibit 9).