This report presents a Conjoint Analysis aimed at understanding the preferences of respondents for various smartphone attributes. Conjoint analysis is a widely used market research technique to quantify the trade-offs consumers make when choosing between different product attributes. This study was conducted to derive insights into which attributes (brand, screen size, memory, camera, and price) matter the most to consumers when selecting a smartphone.
This conjoint analysis is based on a dataset sourced from an Excel file containing two sheets:
Data: This sheet holds the responses from participants who evaluated 30 different combinations of smartphone attributes. Each combination represents a unique product profile that the respondents rated.
Conjoint: This sheet provides the details of the possible combinations of attributes that were presented to the respondents. These combinations were used to structure the questions in the survey.
In addition, a dataframe named level was created to extract and store the unique attribute details for easy reference during analysis.
library(readxl)
library(conjoint)
# Load the data
file_path <- "C:/Program Files/Smartphone conjoint.xlsx"
data_sheet <- read_excel(file_path, sheet = "Data")
data.df <- data.frame(data_sheet)
conjoint_sheet <- read_excel(file_path, sheet = "Conjoint")
conjoint.df <- data.frame(conjoint_sheet)
# Display first few rows of data
head(data.df)## Obs q1 q2 q3 q4 q5 q6 q7 q8 q9 q10 q11 q12 q13 q14 q15 q16 q17 q18 q19 q20
## 1 1 85 80 74 69 67 88 82 83 75 78 90 89 92 95 80 71 96 75 84 83
## 2 2 84 79 77 67 67 83 80 81 79 76 90 86 92 98 82 75 97 77 88 79
## 3 3 82 79 73 67 72 86 80 82 71 82 87 89 95 94 79 75 93 76 80 82
## 4 4 81 76 74 71 66 93 81 78 78 75 93 93 88 94 77 76 98 79 81 81
## 5 5 88 79 77 72 69 83 79 88 73 76 90 87 93 91 81 76 99 74 83 84
## 6 6 84 78 73 67 68 91 80 86 77 74 94 93 90 98 79 75 97 73 85 81
## q21 q22 q23 q24 q25 q26 q27 q28 q29 q30
## 1 90 75 85 75 89 76 86 76 90 87
## 2 92 73 85 73 88 72 85 79 85 89
## 3 85 75 85 73 87 80 87 80 87 92
## 4 89 71 82 76 86 73 86 71 86 85
## 5 92 75 85 78 89 78 86 77 93 90
## 6 95 71 81 74 90 77 84 75 93 84head(conjoint.df)## attribute.1 attribute.2 attribute.3
## 1 Apple 4.5 32G
## 2 Samsung 4.5 32G
## 3 Other global brands (LG, Sony, Nokia, etc.) 4.5 32G
## 4 Local or unknown brands 4.5 32G
## 5 Local or unknown brands 3.5 32G
## 6 Apple 4.5 32G
## attribute.4 attribute.5 Brand Screen.size Memory Camera Price
## 1 8Mp 599 1 2 1 1 4
## 2 8Mp 599 2 2 1 1 4
## 3 8Mp 499 3 2 1 1 3
## 4 8Mp 399 4 2 1 1 2
## 5 8Mp 299 4 1 1 1 1
## 6 12Mp 699 1 2 1 2 5# Creating a unique attribute table(without combinations)
level <- c("apple","samsung","other global brands","local brands","3.5","4,5","5.5","32G","64G","128G","8Mp","12Mp","$299","$399","$499","$599","$699","$799","$899")
level.df <- data.frame(level)Conjoint Analysis Model Utilities Estimation To assess the relative value of each attribute, we calculate the utilities (also known as part-worths) for each attribute level. Higher utility values indicate stronger preferences. The table below presents the utility estimates for the first respondent:
# Calculate utilities of each attributes
caModel(y=data.df[1,2:31],x=conjoint.df[,6:10])##
## Call:
## lm(formula = frml)
##
## Residuals:
## Min 1Q Median 3Q Max
## -5.719 -1.470 0.000 1.086 7.170
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 78.4936 1.2690 61.857 < 2e-16 ***
## factor(x$Brand)1 11.1043 2.6075 4.259 0.000687 ***
## factor(x$Brand)2 5.4078 1.2230 4.422 0.000495 ***
## factor(x$Brand)3 -5.6473 1.4379 -3.928 0.001344 **
## factor(x$Screen.size)1 -9.5034 3.2873 -2.891 0.011198 *
## factor(x$Screen.size)2 4.1962 1.9979 2.100 0.053030 .
## factor(x$Memory)1 -4.4362 1.9453 -2.280 0.037617 *
## factor(x$Memory)2 2.0819 1.2292 1.694 0.110988
## factor(x$Camera)1 -1.3950 1.0066 -1.386 0.186037
## factor(x$Price)1 14.7058 8.1348 1.808 0.090738 .
## factor(x$Price)2 4.7966 3.7500 1.279 0.220297
## factor(x$Price)3 0.1251 2.3689 0.053 0.958587
## factor(x$Price)4 -1.3779 1.7313 -0.796 0.438508
## factor(x$Price)5 -3.8557 2.4762 -1.557 0.140297
## factor(x$Price)6 -5.7394 4.1225 -1.392 0.184154
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 3.618 on 15 degrees of freedom
## Multiple R-squared: 0.8863, Adjusted R-squared: 0.7801
## F-statistic: 8.35 on 14 and 15 DF, p-value: 0.0001008Interpretation: The results show significant positive utility for brands like Apple and Samsung, while local brands carry negative utility. In terms of screen size, larger screens (e.g., 4.5” and 5.5”) are preferred, with 5.5 inches having the highest utility. Lower-priced options, especially $299, also show significant positive utility, suggesting price sensitivity.
Overall Model Utilities The following table presents the average utilities across all respondents for each attribute level:
# For full model
Conjoint(data.df[,2:31], conjoint.df[,6:10], z=level.df)##
## Call:
## lm(formula = frml)
##
## Residuals:
## Min 1Q Median 3Q Max
## -10,5591 -2,8960 -0,0404 2,8400 11,2535
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 78,37946 0,11143 703,368 <2e-16 ***
## factor(x$Brand)1 11,21894 0,22898 48,995 <2e-16 ***
## factor(x$Brand)2 5,30712 0,10740 49,414 <2e-16 ***
## factor(x$Brand)3 -5,69082 0,12627 -45,069 <2e-16 ***
## factor(x$Screen.size)1 -9,60015 0,28868 -33,256 <2e-16 ***
## factor(x$Screen.size)2 4,20635 0,17545 23,974 <2e-16 ***
## factor(x$Memory)1 -4,43315 0,17083 -25,951 <2e-16 ***
## factor(x$Memory)2 2,11200 0,10795 19,565 <2e-16 ***
## factor(x$Camera)1 -1,41446 0,08839 -16,002 <2e-16 ***
## factor(x$Price)1 14,67022 0,71437 20,536 <2e-16 ***
## factor(x$Price)2 4,64628 0,32931 14,109 <2e-16 ***
## factor(x$Price)3 0,26045 0,20803 1,252 0,211
## factor(x$Price)4 -1,33590 0,15203 -8,787 <2e-16 ***
## factor(x$Price)5 -3,86032 0,21745 -17,753 <2e-16 ***
## factor(x$Price)6 -5,81291 0,36202 -16,057 <2e-16 ***
## ---
## Signif. codes: 0 '***' 0,001 '**' 0,01 '*' 0,05 '.' 0,1 ' ' 1
##
## Residual standard error: 3,891 on 4485 degrees of freedom
## Multiple R-squared: 0,7727, Adjusted R-squared: 0,772
## F-statistic: 1089 on 14 and 4485 DF, p-value: < 2,2e-16## [1] "Part worths (utilities) of levels (model parameters for whole sample):"
## levnms utls
## 1 intercept 78,3795
## 2 apple 11,2189
## 3 samsung 5,3071
## 4 other global brands -5,6908
## 5 local brands -10,8352
## 6 3.5 -9,6002
## 7 4,5 4,2063
## 8 5.5 5,3938
## 9 32G -4,4331
## 10 64G 2,112
## 11 128G 2,3211
## 12 8Mp -1,4145
## 13 12Mp 1,4145
## 14 $299 14,6702
## 15 $399 4,6463
## 16 $499 0,2604
## 17 $599 -1,3359
## 18 $699 -3,8603
## 19 $799 -5,8129
## 20 $899 -8,5678
## [1] "Average importance of factors (attributes):"
## [1] 31,56 21,44 10,85 3,93 32,22
## [1] Sum of average importance: 100
## [1] "Chart of average factors importance"Apple and Samsung are the most preferred brands, while local brands have significantly lower utility. Larger screen sizes (4.5” and 5.5”) are more desirable. Consumers exhibit a strong preference for lower price points, particularly $299 and $399. Attribute Importance Next, we evaluate the importance of each attribute in the decision-making process. Attribute importance represents the percentage contribution of each attribute to the overall decision.
# Calculate attribute importance
caImportance(data.df[,2:31], conjoint.df[,6:10])## [1] 31.56 21.44 10.85 3.93 32.22The output above shows the importance of each attribute:
Price (32.22%) is the most influential factor in decision-making. Brand (31.56%) is nearly as important, indicating that consumers prioritize brand recognition along with price. Screen Size (21.44%) is the third most important, while Memory (10.85%) and Camera (3.93%) have lower influence.
Recommendations: Brands that want to capture a larger market share should focus on optimizing pricing strategies while maintaining a strong brand image. Offering smartphones at competitive prices while emphasizing screen size could yield better consumer response.
Consumer Segmentation To identify distinct consumer segments based on preferences, we applied a K-means clustering algorithm. This technique helps group consumers with similar preferences into clusters.
# Perform segmentation
caSegmentation(data.df[,2:31],conjoint.df[,6:10],c=3)The analysis grouped respondents into three clusters:
Cluster 1: Price-sensitive consumers who value lower-priced models and are willing to compromise on brand or other features.
Cluster 2: Consumers with moderate preferences across attributes, leaning slightly towards larger screen sizes and mid-range prices.
Cluster 3: Brand-loyal consumers who prioritize premium brands like Apple and Samsung over price. Actionable Insights: Smartphone manufacturers should consider targeting these segments with tailored marketing strategies. For example, premium brands could focus on Cluster 3 by highlighting brand prestige, while Cluster 1 could be targeted with budget-friendly models.
The conjoint analysis reveals that price and brand are the most critical factors in smartphone purchasing decisions, followed by screen size. This suggests that competitive pricing and a strong brand presence are key strategies for smartphone companies. Manufacturers targeting different consumer segments can adjust their product offerings to match the preferences revealed in this analysis.