To facilitate the readers^, understanding, you may use two patterns: 

1) highlight the main results in a single sentence placed in the very beginning of a paragraph (Example 1)

2) report the result right after the location of the results (Example 2). 

At the beginning, the topic sentence contains the main results, you then select related data to support it. Steps are:

• Analyze data obtained and find related data.

• Find the meaningful results (shown in the trends and/or irregularities in the data).

• Synthesize the results into a pattern and fit them into the answers to the research questions raised in the Introduction section.

• Use the TS-SS structure by beginning the data presentation with a topic sentence which contain the findings or the answers, and then writing supporting sentences of examples and figures.

• Avoid duplicating the data you showed in tables and/or figures when describing the tables and figures. Instead, you may tell the meaning of the data, which may not show in the table or figure. For instance, you may write “Warming increased methane emissions by a factor of 3.4 (p<0.05)“ instead of "Warming significantly increased methane emissions by a factor of 3.4” 

Example 1 :

The wide gap between the women's expressed opinions about infant feeding and what they actually practiced is further \ underscored by the data in table 8. Contrary to their actual practice as indicated in table 6, 50% of these mothers believed that infants aged three to six months thrived best on breast-feeding alone, while about 35% shared the view that a combination of  breast- and bottle-feeding was the right answer (table 8). There was some suggestive evidence from the studies that bottle feeding was not very strongly or extensively promoted by the hospital/clinic staff attending these mothers, although quite a number of the women (22%) claimed that they received free milk samples for an ill-defined period after delivery of their babies (tables 9 and 10). 

Example 2:

The results are shown in Table 3. Remarkably, every classifer achieves significantly higher than random guessing (AUC- j )C=0.5). The results confirm the feasibility of forecasting severe depression based on individual histories. LSTM-RNN (all) tperforms SVM (all) regardless of n values. From the results, we confirm that LSTM-RNN appropriately processes time series ormation to forecast severe depression. Compared by feature sets, all shows higher AUC-ROC than severe only. Therefore, » confirm that fine-grained information such as moods at different parts ... These results are inconsistent with previous studies it have shown a correlation between sleeping irregularity and negative moods [37]. We consider the absence of contribution these features is caused by building a single predictive model among all users in the dataset. Some users might suffer m severe depression without attending hospitals or taking medicines. This result thus implies that stratifying users to create Htiple predictive models for each group might improve forecasting performance.

以下内容结合四六级考研写作