4.1.
Descriptive statistics and visualizations
Understanding the main findings and successfully presenting the findings to stakeholders depend
on these statistical metrics and graphic depictions.
In order to have a thorough comprehension of the system's performance, we computed multiple
summary statistics. These statistics offer a quantitative summary of the several facets of the
chatbot framework. We determined how long it typically took the chatbot to reply to questions
from customers. This metric aids in evaluating how well the system responds quickly. By
contrasting the chatbot's responses with a list of predetermined right answers, we were also able
to determine how accurate they were. The accuracy of the information provided by the system is
shown by this statistic.
We used a variety of visualizations to further examine and present the results. These visual aids
facilitate the identification of patterns and trends and provide a more intuitive comprehension of
the data.
A visual depiction of the frequency distribution of word counts in the text data is given by the bar
graph below. The word count range is represented by the x-axis, and the frequency or count of
occurrences for each word count range is represented by the y-axis.
Figure …. : Bar graph showing distribution of word counts in questions
The distribution of word counts can be inferred from the bar graph's shape. It can help us
comprehend the average text size or length. The majority of the text data has shorter word counts
if the bar graph is slanted to the left. Longer word counts are indicated if it is tilted to the right.
The histogram's various peaks or modes indicate that the text data may contain different
groupings or patterns. The bar graph in this instance is slanted to the left, indicating that shorter
questions are more prevalent than longer ones.
The sentiment polarity or sentiment scores connected to each piece of text data are displayed in
the sentiment analysis plot below. It aids in our comprehension of the text's overall attitude or
emotive tone.
Figure … : Sentiment analysis plot showing the sentiment polarity or sentiment scores
associated with each piece of text data.
Sentiment scores that are positive represent feelings that are positive, and those that are negative
represent feelings that are negative. We can spot patterns or trends in the stated sentiment by
examining the plot, which can show how sentiment is distributed throughout the dataset.
Additionally, we can examine how sentiment has changed over time or compare sentiment
ratings among various dataset subsets.