"Probability distributions have personalities too."
Aoi suddenly said.
"Personalities?" Yuki asked curiously.
"It's interesting to personify them. Uniform distribution is an equalitarian, normal distribution prefers moderation, etc."
Riku laughed. "That metaphor again."
At that moment, Mira quietly approached and showed graphs of various distributions drawn in a notebook.
"This is..." Yuki stared.
"Various probability distributions," Aoi began explaining. "Each has different properties."
Aoi pointed to the first diagram.
"Uniform distribution. Like dice, all outcomes equally probable. Maximum entropy."
"A fair distribution," Riku said.
"Yes. But rare in nature. Most phenomena have bias."
Mira pointed to another graph. A bell-shaped curve.
"Normal distribution," Aoi continued. "Most common in nature. Height, weight, measurement errors. Concentrated at center, rare at extremes."
"Why is normal distribution so common?" Yuki asked.
"Central limit theorem. Sum many independent variables, and you approach normal distribution."
Riku pondered. "It's strange that complex phenomenon results become simple distributions."
"That's the beauty of mathematics," Aoi said.
Mira showed the next graph. A distribution skewed to one side.
"Exponential distribution," Aoi explained. "Often used for waiting times and lifetimes. Models 'time until next occurrence.'"
Yuki wrote in their notebook. "I've heard it has memorylessness."
"Excellent," Aoi was impressed. "Even knowing the past, future probability doesn't change. Used for bus waiting."
"But actual buses have schedules, so it's different," Riku pointed out.
"Sharp. Models are approximations of reality. Not perfect, but useful."
Mira opened another page. Discrete bar graphs.
"Binomial distribution," Aoi said. "Multiple coin tosses. Distribution of success counts."
"Like Pokemon games?" Riku associated.
"Item drop rates are exactly binomial distribution."
Yuki asked. "When choosing distributions, how do you decide?"
"Data properties and problem model," Aoi answered. "Continuous or discrete, finite or infinite range, independence present?"
Mira wrote a note. "Distribution is a language for modeling randomness"
"Yes. Probability distributions are languages for describing randomness."
Riku compared multiple graphs. "Do all these have different entropy?"
"Of course. Even on the same support, different distributions have different entropy."
Aoi showed calculation examples.
"Uniform distribution has maximum entropy. The more biased, the lower the entropy."
"If completely determined, zero," Yuki understood.
"Exactly. Delta distribution, with all probability concentrated at one point, has zero entropy."
Mira drew a new figure. A complex multimodal distribution.
"What's this?" Riku asked.
"Mixture distribution. Multiple groups mixed together," Aoi explained. "For example, combining male and female height distributions shows two peaks."
Yuki was impressed. "Looking at distributions reveals data background."
"Exactly. In data science, understanding distribution shape is the first step."
Riku looked out the window. "Do leaf sizes also follow some distribution?"
"Probably. Observe and collect data to find out."
Mira left a final note. "Every distribution tells a story"
"Every distribution tells a story," Yuki translated.
Aoi nodded. "Probability distributions are windows for understanding the world."
The three quietly imagined the invisible world of distributions.