![](bg8.png)
Table 3: Forecast results for Target 3.2, the year 2030 and
the country Algeria.
Indicator Age/Sex Initial Target Forecast Result
3.2.1 1Y/F 20.2 <=25 16.94 Met
3.2.1 1Y/M 22.9 <=25 20.82 Met
3.2.1 5Y/F 23.7 <=25 19.89 Met
3.2.1 5Y/F 26.6 <=25 24.13 Met
3.2.2 1Month/FM 15 <=12 13.75 Not Met
Figure 5: Vitalising SDG attainment using D3.js.
each indicator The sixth and seventh columns, “Initial
Value” and “Prediction”, gives the mortality value per
1000 live births in 2015, and the predicted value in
2030. The final SDG attainment prediction result is
given in the last column. For Target 3.2 to be attained
(met), the value associated with each indicator (time
series) must meet its threshold (at or below the rele-
vant threshold in this case). Unfortunately, in this ex-
ample, all of the indicators meet the required thresh-
old before 2030 except 3.2.2. Thus it is concluded
that Target 3.2 will not be attained.
The SDG-TTF framework includes a visualisation
mechanism, as indicated in Figure 4. This was imple-
mented using D3.js (Bostock et al., 2011). The vi-
sualisation allows users to: (i) track the progress of
different goals over a given time frame, and (ii) trace
the achievement of individual bottom level indicators
in an interactive manner. An example of such visual-
isations is given in Figure 5 using the case study pre-
sented above. From the figure it can be seen that using
the visualisation it is easy to identify goal attainment
(or non-attainment as in this case). Nodes coloured
in green highlight indicators/targets/goals that will be
attained on time. Nodes coloured in red highlight in-
dicators/targets/goals that will not be attained on time.
For a more detailed analysis of why a goal is not at-
taining the relevant country table can give a better ex-
planation.
8 CONCLUSION
In this paper we have presented the SDG-TTF attain-
ment prediction framework. Unlike previous frame-
works directed at SDG attainment prediction the
SDG-TTF framework takes into consideration both
inter- and intra-geographic entity (county, region)
causal correlation. The intuition was that individ-
ual SDG indicators should not be considered in isola-
tion because inspection of the indicators demonstrates
clear potential for causal relationships with respect to
other indicators for the entity in question and with re-
spect to indicators in neighbouring entities. The eval-
uation of the framework demonstrates that more ro-
bust SDG attainment predictions using SDF-TTF can
be made. For future work the authors intend to inves-
tigate further alternative causal relationship discovery
mechanisms; and to give further consideration of the
parameter k, the number of time series to be included
when building the multi-variate time series prediction
models central to the SDG-TTF framework. Finally
the authors intend to use the framework to investigate
the effect on SDG attainment in presence of natural
disasters, such as the Covid-19 pandemic, which oc-
cur for short periods of time but might have a signifi-
cant impact on SDG attainment prediction.
REFERENCES
Alam, N. and Rudin, C. (2015). Robust Nonparametric
Testing for Causal Inference in Observational Studies.
Optimization Online, Dec, pages 1–39.
Alharbi, Y., Arribas-Bel, D., and Coenen, F. (2019).
Sustainable development goal attainment prediction:
A Hierarchical Framework using Time Series Mod-
elling. IC3K 2019, 1:297–304.
Alharbi, Y., Coenen, F., and Arribas-Bel, D. (2020). Sus-
tainable development goal relational modelling: In-
troducing the SDG-CAP methodology. In DAWAK),
volume 12393 LNCS, pages 183–196.
Athanasopoulos, G., Ahmed, R. A., and Hyndman, R. J.
(2009). Hierarchical forecasts for Australian domestic
tourism. IFJ, 25(1):146–166.
Bostock, M., Ogievetsky, V., and Heer, J. (2011). D3 data-
driven documents. IEEE TVCG, 17.
Chen, K., Zhou, Y., and Dai, F. (2015). A LSTM-based
method for stock returns prediction: A case study of
China stock market. In IEEE Big Data 2015. IEEE.
De Gooijer, J. G. and Hyndman, R. J. (2006). 25 Years of
Time Series Forecasting. IFJ, 22(3):443–473.
Deschamps, E. (2004). The impact of institutional change
on forecast accuracy: A case study of budget forecast-
ing in Washington State. IFJ, 20(4):647–657.
D
¨
orgo, G., Sebesty
´
en, V., and Abonyi, J. (2018). Evaluating
the interconnectedness of the sustainable development
goals based on the causality analysis of sustainability
indicators. Sustainability (Switzerland), 10(10):3766.
Epprecht, C., Guegan, D., and Veiga,
´
A. (2013). Com-
paring variable selection techniques for linear regres-
sion: LASSO and Autometrics. Centre d’
´
economie de
la Sorbonne.
Frey, B. B. (2018). Pearson Correlation Coefficient. In The
SAGE Encyclopedia of Educational Research, Mea-
surement, and Evaluation, pages 1–4. Springer.
DeLTA 2021 - 2nd International Conference on Deep Learning Theory and Applications
130