The growing popularity of multimedia video streaming services, combined with thernincreased prevalence of smartphones, has resulted in impressive demand growth andrna signi cant need for quality of service (QoS). As a result, making video streamingrnservices available to users in a cost-e ective manner without sacri cing quality is arnsigni cant challenge. In Ethiopia, the sole telecom services provider, Ethio Telecomrn(ET), has launched several network expansion projects in recent years to providernquality services. However, there are still customer complaints from di erent partsrnof the city.rnThis thesis intends to study the spatio-temporal correlation analysis among Qual-rnity of Experience (QoE) and QoS parameters for LTE video streaming service andrnindicate measurement variations. The objective data collection has been collectedrnusing Nemo Handy, Opensignal, and Cloudrnare speed tests. The subjective percep-rntions of video streaming users were gathered using a contextually formulated surveyrnquestionnaire via a eld experiment campaign including 300 di erent sample usersrnwho watched a single YouTube video session. This study is conducted in six di erentrnlocations considering the application, network, and physical layer quality metrics,rnand we analyze the data using Statistical Package for the Social Sciences(SPSS),rnActix Analyzer, and Excel.rnIn all study locations, in particular during the evening and nighttime periods, thernmeasured latency and jitter are not suitable for video streaming services. Duringrnthese periods, the Mean Opinion Scores (MOS) are less than 3.3 and on average,rna MOS result of 3.9 is obtained. According to the correlation analysis results, thernhighest correlation magnitude is achieved between MOS and playback with a strongrnpositive and statistically signi cant value of (r = .746, p < .001). That is, playback isrnthe main factor responsible for the MOS variation. The strongest correlation amongrnapplication and network layer parameters is found between download throughputrnand playback, with values of (r =.711, p < .001). As a result, rather than latencyrnand jitter, download throughput is the main reason for the variation of application layer quality metrics. We found weaker correlation results between network andrnphysical layer quality metrics at Mebrathail and Jemo areas, indicating that thernnetwork layer quality matrices are inrnuenced by the increase in the number of usersrnand other factors, rather than RSRP and SINR.