Prediction
A prediction is a specific statement of what should happen in an experiment if the hypothesis is correct.
Definition
A prediction is what you expect to see when you run your experiment; it comes from the hypothesis and tells you what result to look for. The hypothesis explains why, while the prediction states what will be measured, and comparing actual data to the prediction is how the hypothesis is evaluated. Logically, a prediction is a deductive consequence: if the hypothesis and the setup are as described, a particular measurable result should follow. Precise predictions state the expected magnitude and the uncertainty within which a result still counts as agreement; a hypothesis whose predictions fail under fair conditions is falsified, and one whose predictions succeed is supported but not proven.
Example
If the hypothesis is that warm water dissolves salt faster, the prediction is "the salt in the warm cup will disappear first." For plants and light, a prediction with a number: "after 14 days, seedlings under the lamp will average at least $3\ \text{cm}$ taller than seedlings in the shaded box," and means of $9.2\ \text{cm}$ versus $5.4\ \text{cm}$ confirm it. From energy conservation, a cart released $0.40\ \text{m}$ up a frictionless ramp should reach $v = \sqrt{2gh} = \sqrt{2 \times 9.8 \times 0.40} \approx 2.8\ \text{m/s}$; a group predicts slightly less because of friction, and a five-trial mean of $2.6\ \text{m/s}$ lands in the expected band.
Key Insight
Make the prediction before you start, because deciding what counts as success after seeing the data is not fair to the experiment. Include a number or a clear direction (taller, faster, more) so the result cannot be argued both ways. The riskier the prediction, the more it teaches: a hypothesis that predicts a specific value and gets it earns far more credibility than one that predicted "some change".